No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Large-Scale Evaluation of ANTs and FreeSurfer Cortical Thickness Measurements.
Abstract
Many studies of the human brain have explored the relationship between cortical thickness and cognition, phenotype, or disease. Due to the subjectivity and time requirements in manual measurement of cortical thickness, scientists have relied on robust software tools for automation which facilitate the testing and refinement of neuroscientific hypotheses. The most widely used tool for cortical thickness studies is the publicly available, surface-based FreeSurfer package. Critical to the adoption of such tools is a demonstration of their reproducibility, validity, and the documentation of specific implementations that are robust across large, diverse imaging datasets. To this end, we have developed the automated, volume-based Advanced Normalization Tools (ANTs) cortical thickness pipeline comprising well-vetted components such as SyGN (multivariate template construction), SyN (image registration), N4 (bias correction), Atropos (n-tissue segmentation), and DiReCT (cortical thickness estimation). In this work, we have conducted the largest evaluation of automated cortical thickness measures in publicly available data, comparing FreeSurfer and ANTs measures computed on 1205 images from four open data sets (IXI, MMRR, NKI, and OASIS), with parcellation based on the recently proposed Desikan-Killiany-Tourville (DKT) cortical labeling protocol. We found good scan-rescan repeatability with both FreeSurfer and ANTs measures. Given that such assessments of precision do not necessarily reflect accuracy or ability to make statistical inferences, we further tested the neurobiological validity of these approaches by evaluating thickness-based prediction of age and gender. ANTs is shown to have a higher predictive performance than FreeSurfer for both of these measures. In promotion of open science, we make all of our scripts, data, and results publicly available which complements the use of open image data sets and the open source availability of the proposed ANTs cortical thickness pipeline.
... Structurally, OUD patients show lower gray matter volumes and/or lower cortical thickness compared to non-dependent controls, particularly in the prefrontal and temporal cortices [29][30][31]. Cortical thickness is a well-established and clinically relevant metric of brain structural integrity [32][33][34] and has been linked to OUD severity and treatment outcomes [29,35]. However, research on the effect of abstinence and treatment on brain morphology remains limited and has yielded inconsistent results [35][36][37]. ...
... High-resolution T1-weighted whole-brain images were acquired using the magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence with repetition time/echo time = 1510/ 3.71 ms, field of view = 256 × 192 mm², matrix = 256 × 192, slice thickness/gap = 1/0 mm, 160 slices, effective voxel resolution = 1 × 1 × 1 mm³, and flip angle = 9°. MRI images were processed using the volume-based cortical thickness pipeline implemented in Advanced Normalization Tools (ANTs) [34,44]. Details of the pipeline can be found in Tustison et al. [34]. ...
... MRI images were processed using the volume-based cortical thickness pipeline implemented in Advanced Normalization Tools (ANTs) [34,44]. Details of the pipeline can be found in Tustison et al. [34]. Specifically, N4 bias field correction was performed on the raw structural images to minimize the impact of low-frequency field inhomogeneity [45]. ...
Opioid use disorder (OUD) has been linked to macroscopic structural alterations in the brain. The monthly injectable, extended-release formulation of μ-opioid antagonist naltrexone (XR-NTX) is highly effective in reducing opioid craving and preventing opioid relapse. Here, we investigated the neuroanatomical effects of XR-NTX by examining changes in cortical thickness during treatment for OUD. Forty-seven OUD patients underwent structural magnetic resonance imaging and subjectively rated their opioid craving ≤1 day before (pre-treatment) and 11 ± 3 days after (on-treatment) the first XR-NTX injection. A sample of fifty-six non-OUD individuals completed a single imaging session and served as the comparison group. A publicly available [¹¹C]carfentanil positron emission tomography dataset was used to assess the relationship between changes in cortical thickness and μ-opioid receptor (MOR) binding potential across brain regions. We found that the thickness of the medial prefrontal and anterior cingulate cortices (mPFC/aCC; regions with high MOR binding potential) was comparable between the non-OUD individuals and the OUD patients at pre-treatment. However, among the OUD patients, mPFC/aCC thickness significantly decreased from pre-treatment to on-treatment. A greater reduction in mPFC/aCC thickness was associated with a greater reduction in opioid craving. Taken together, our study suggests XR-NTX-induced cortical thickness reduction in the mPFC/aCC regions in OUD patients. The reduction in thickness does not appear to indicate a restoration to the non-OUD level but rather reflects XR-NTX’s distinct therapeutic impact on an MOR-rich brain structure. Our findings highlight the neuroplastic effects of XR-NTX that may inform the development of novel OUD interventions.
... The lesioned area was manually segmented on the T1w images using ITK-SNAP [14]. The UNC 1-year AAL parcellation atlas [15] was registered to the Visit-1 T1w scan using FLIRT [16] for affine registration, followed by ANTs diffeomorphic registration [17], providing an anatomically matched motor cortex (M1) mask. This mask was used to estimate M1 tissue loss [15] in the lesioned hemisphere M1, which was calculated as Volume (M1 mask-lesion mask)/Volume (M1 mask). ...
... TractSeg was used to segment white matter bundles and generate probabilistic tractograms for bilateral CSTs [23]. The mean of the non-diffusion-weighted images was affinely aligned to the MNI space using ANTs [17]. The resulting transform was applied to the CSD-based fiber orientation distribution (FOD) maps, which were included as part of the TractSeg workflow. ...
Background: Perinatal brain injury is a leading cause of developmental disabilities, including cerebral palsy. However, further work is needed to understand early brain development in the presence of brain injury. In this case report, we examine the longitudinal neuromotor development of a term infant following a significant loss of right-hemispheric brain tissue due to a unilateral ischemic stroke. Our analysis focuses on the integrity and development of the corticospinal tract (CST) from the lesioned hemisphere. This case provides a unique opportunity to evaluate CST development after loss of the majority of the motor cortex. Methods: Evaluations were conducted when the infant was 4 (Visit-1), 18 (Visit 2), and 25 (Visit 3) months old. Assessments included magnetic resonance imaging (MRI) to characterize the lesion and quantify CST structural integrity, single-pulse transcranial magnetic stimulation (spTMS) to evaluate CST functional circuitry, and neuromotor assessments. Results: At Visit 1, bilateral CSTs were identified through diffusion-weighted MRI (dMRI) despite an estimated loss of 92.7% (7.3% retained) of age-typical motor cortex from the right hemisphere. Both hemispheres exhibited bilateral motor-evoked potential in response to stimulation with spTMS, which remained when reassessed at Visits 2 and 3. Longitudinal MRI showed distinct developmental trajectories of CST integrity in each hemisphere, with the lesioned hemisphere exhibiting initial increases in integrity between Visits 1 and 2 followed by a decrease in integrity between Visits 2 and 3. The non-lesioned hemisphere showed increased integrity from Visit 1 to Visit 2, which remained stable at Visit 3. Motor assessments at all visits indicated a high risk of cerebral palsy. Conclusions: This report highlights the utility of MRI and spTMS in studying neuromotor development. The findings reveal preserved functional bilateral CST circuitry despite majority loss of the right-hemispheric motor cortex as well as distinct developmental trajectories in CST integrity between hemispheres. These results underscore the potential for neural plasticity after perinatal brain injury. Clinical Trials Registration: NCT05013736.
... Women were ineligible if they were: (1) not able to read or speak English, (2) 16 with right and left hemispheric values averaged as previously described. 17 We then estimated cortical thickness from these segmentations using registration-based cortical thickness 18 within the regions of AD-pattern neurodegeneration. These regions included the entorhinal cortex as the earliest region to be involved in AD-related neurodegeneration on MRI, 19 ...
... We thank Avid Radiopharmaceuticals, Eli Lilly for their hep with the production of 18 ...
INTRODUCTION
Premenopausal bilateral oophorectomy (PBO) before the age of 46 years is associated with an increased risk of dementia. We investigated the long‐term effects of PBO performed before age 50 years on amyloid beta (Aβ), tau, and neurodegeneration imaging biomarkers of Alzheimer's disease (AD).
METHODS
Mayo Clinic Cohort Study of Oophorectomy and Aging‐2 participants were divided into early PBO (< 46 years; n = 61), and late PBO (46–49 years; n = 51) groups and were compared to referent women who did not undergo PBO (n = 119).
RESULTS
Early PBO was associated with thinner entorhinal cortex (p = 0.014), higher tau load at higher levels of Aβ load (Pp = 0.005), higher Aβ load (p = 0.026), and smaller temporal lobe cortical thickness (p = 0.022), only at older ages compared to the referent group.
DISCUSSION
PBO before the age of 46 years is associated with entorhinal cortex thinning, elevated tau at higher Aβ levels, along with an AD‐like pattern of atrophy at older ages.
CLINICAL TRIALS REGISTRATION
NCT03821857 sex‐specific effects of endocrine disruption on aging and AD.
Highlights
Premenopausal bilateral oophorectomy (PBO) before the ages of 46 (early PBO) years and ages 46 to 49 (late PBO) years was studied.
Early PBO was associated with reduced entorhinal cortex thickness later in life.
Early PBO was associated with greater amyloid beta (Aβ) load at older ages.
Early PBO was associated with greater Alzheimer's disease pattern of atrophy at older ages.
Early PBO was associated with higher tau load at higher Aβ levels.
... We evaluated the performance of the proposed ReMiND models with respect to volumetric features extracted using the ANTs pipeline ( Tustison et al., 2014) and Free-Surfer's pipeline ( Fischl, 2012). The version of ANTs is 0.3.4 for ANTsPy. ...
... Under each experiment setting (P and PF), ReMiND models had the lowest error rates compared with the two comparator methods, and the Naive method had the highest error rates. Furthermore, ANTs-based error rates were lower across all methods and experimental settings compared to FreeSurfer as expected based on previous work ( Tustison et al., 2014). ...
Missing data is a significant challenge in medical research. In longitudinal studies of Alzheimer’s disease (AD) where structural magnetic resonance imaging (MRI) is collected from individuals at multiple time points, participants may miss a study visit or drop out. Additionally, technical issues such as participant motion in the scanner may result in unusable imaging data at designated visits. Such missing data may hinder the development of high-quality imaging-based biomarkers. To address the problem of missing MRI data in studies of AD, we introduced a novel 3D diffusion model specifically designed for imputing missing structural MRI (Recovery of Missing Neuroimaging using Diffusion models (ReMiND)). The model generates a whole-brain image conditional on a single structural MRI observed at a past visit or conditional on one past and one future observed structural MRI relative to the missing observation. The performance of models was compared with two alternative imputation approaches: forward filling and image generation using variational autoencoders. Experimental results show that our method can generate 3D structural MRI with high similarity to ground-truth images at designated visits. Furthermore, images generated using ReMiND show relatively lower differences in volume estimation between the imputed and observed images compared to images generated by forward filling or autoencoders. Additionally, ReMiND provides more accurate estimated rates of atrophy over time in important anatomical brain regions than the two comparator methods. Our 3D diffusion model can impute missing structural MRI data at a single designated visit and outperforms alternative methods for imputing whole-brain images that are missing from longitudinal trajectories.
... The registration process begins with co-registering the highest quality images from each modality to the synthetic T1w derivative (Supplementary Figure 2). We then create a transformation between this synthesized T1w image and the MNI-152 T1w template using a combination of rigid, affine, and non-linear transformations implemented through ITK 31 and ANTs 35,36 , with brain masks defining the registration sampling region (Supplementary Figure 3). This single . ...
Predicting long-term functional outcomes for individuals with stroke is a significant challenge. Solving this challenge will open new opportunities for improving stroke management by informing acute interventions and guiding personalized rehabilitation strategies. The location of the stroke is a key predictor of outcomes, yet no clinically deployed tools incorporate lesion location information for outcome prognostication. This study responds to this critical need by introducing a fully automated, three-stage neuroimaging processing and machine learning pipeline that predicts personalized outcomes from clinical imaging in adult ischemic stroke patients. In the first stage, our system automatically processes raw DICOM inputs, registers the brain to a standard template, and uses deep learning models to segment the stroke lesion. In the second stage, lesion location and automatically derived network features are input into statistical models trained to predict long-term impairments from a large independent cohort of lesion patients. In the third stage, a structured PDF report is generated using a large language model that describes the stroke's location, the arterial distribution, and personalized prognostic information. We demonstrate the viability of this approach in a proof-of-concept application predicting select cognitive outcomes in a stroke cohort. Brain-behavior models were pre-trained to predict chronic impairment on 28 different cognitive outcomes in a large cohort of patients with focal brain lesions (N=604). The automated pipeline used these models to predict outcomes from clinically acquired MRIs in an independent ischemic stroke cohort (N=153). Starting from raw clinical DICOM images, we show that our pipeline can generate outcome predictions for individual patients in less than 3 minutes with 96% concordance relative to methods requiring manual processing. We also show that prediction accuracy is enhanced using models that incorporate lesion location, lesion-associated network information, and demographics. Our results provide a strong proof-of-concept and lay the groundwork for developing imaging-based clinical tools for stroke outcome prognostication.
... Existing automated methods, such as cortical thickness measurements [10][11][12] and voxel-based morphometry, 13 aim to provide objective assessments by segmenting the boundary between gray matter and white matter. However, these methods often face challenges due to the subtle intensity differences at this interface, resulting in variability across imaging protocols and vendors. ...
Background and Purpose
Brain atrophy, characterized by sulcal widening and ventricular enlargement, is a hallmark of neurodegenerative diseases such as Alzheimer’s disease. Visual assessments are subjective and variable, while automated methods struggle with subtle intensity differences and standardization, highlighting limitations in both approaches. This study aimed to develop and evaluate a novel method focusing on cerebrospinal fluid (CSF) regions by assessing segmentation accuracy, detecting stage-specific atrophy patterns, and testing generalizability to unstandardized datasets.
Methods
We utilized T1-weighted magnetic resonance imaging data from 3,315 participants from Samsung Medical Center and 1,439 participants from other hospitals. Segmentation accuracy was evaluated using the Dice similarity coefficient (DSC), and W-scores were calculated for each region of interest (ROI) to assess stage-specific atrophy patterns.
Results
The segmentation demonstrated high accuracy, with average DSC values exceeding 0.9 for ventricular and hippocampal regions and above 0.8 for cortical regions. Significant differences in W-scores were observed across cognitive stages (cognitively unimpaired, mild cognitive impairment, dementia of Alzheimer’s type) for all ROIs (all, p<0.05). Similar trends were observed in the images from other hospitals, confirming the algorithm’s generalizability to datasets without prior standardization.
Conclusions
This study demonstrates the robustness and clinical applicability of a novel CSF-focused segmentation method for assessing brain atrophy. The method provides a scalable and objective framework for evaluating structural changes across cognitive stages and holds potential for broader application in neurodegenerative disease research and clinical practice.
... Having identified the best performing lesion filling method, further evaluations centers on the influence of lesion filling on cortical thickness measurements. Cortical thickness assessments are performed on sixty-five patients from the test set, both before and after lesion filling, using five methods: ANTs [37], ANTsPyNet [38], FreeSurfer [39], FastSurfer [40] and DL+DiReCT [26]. FreeSurfer calculates cortical thickness by modeling the cortical band as a surface mesh. ...
Cortical thickness measurements from magnetic resonance imaging, an important biomarker in many neurodegenerative and neurological disorders, are derived by many tools from an initial voxel-wise tissue segmentation. White matter (WM) hypointensities in T1-weighted imaging, such as those arising from multiple sclerosis or small vessel disease, are known to affect the output of brain segmentation methods and therefore bias cortical thickness measurements. These effects are well-documented among traditional brain segmentation tools but have not been studied extensively in tools based on deep-learning segmentations, which promise to be more robust. In this paper, we explore the potential of deep learning to enhance the accuracy and efficiency of cortical thickness measurement in the presence of WM lesions, using a high-quality lesion filling algorithm leveraging denoising diffusion networks. A pseudo-3D U-Net architecture trained on the OASIS dataset to generate synthetic healthy tissue, conditioned on binary lesion masks derived from the MSSEG dataset, allows realistic removal of white matter lesions in multiple sclerosis patients. By applying morphometry methods to patient images before and after lesion filling, we analysed robustness of global and regional cortical thickness measurements in the presence of white matter lesions. Methods based on a deep learning-based segmentation of the brain (Fastsurfer, DL+DiReCT, ANTsPyNet) exhibited greater robustness than those using classical segmentation methods (Freesurfer, ANTs).
... DBM maps were derived from each participant's T1-weighted MRI image at each time point using the Advanced Normalization Tools (ANTs) Longitudinal Cortical Thickness pipeline (see Deformationbased morphometry (DBM) processing in Supplementary Material). 30,31 Quality control was done by visual inspection of the resultant DBM maps. In total, 158 subjects with AD (19%), 40 FHAD subjects (18%) and 67 HC (18%) were excluded mostly due to inaccurate gray matter segmentation (see Table 1 for the remaining number of subjects at each time point). ...
Alzheimer's disease (AD) is associated with presymptomatic changes in brain morphometry and accumulation of abnormal tau and amyloid-beta pathology. Studying the development of brain changes prior to symptoms onset may lead to early diagnostic biomarkers and a better understanding of AD pathophysiology. AD pathology is thought to arise from a combination of protein accumulation and spreading via neural connections, but how these processes influence brain atrophy progression in the presymptomatic phases remains unclear. Individuals with a family history of AD (FHAD) have an elevated risk of AD, providing an opportunity to study the presymptomatic phase.
Here we used structural MRI from three databases (Alzheimer's Disease Neuroimaging Initiative, Pre-symptomatic Evaluation of Experimental or Novel Treatments for Alzheimer Disease and Montreal Adult Lifespan Study) to map atrophy progression in FHAD and AD and assess the constraining effects of structural connectivity on atrophy progression. Cross-sectional and longitudinal data up to four years were used to perform atrophy progression analysis in FHAD and AD compared to controls. Positron emission tomography radiotracers were also used to quantify the distribution of abnormal tau and amyloid-beta protein isoforms at baseline. We first derived cortical atrophy progression maps using deformation-based morphometry from 153 FHAD, 156 AD, and 116 controls with similar age, education, and sex at baseline. We next examined the spatial relationship between atrophy progression and spatial patterns of tau aggregates and amyloid-beta plaques deposition, structural connectivity, and neurotransmitter receptor and transporter distributions.
Our results show that there were similar patterns of atrophy progression in FHAD and AD, notably in the cingulate, temporal, and parietal cortices, with more widespread and severe atrophy in AD. Both tau and amyloid-beta pathology tended to accumulate in regions that were structurally connected in FHAD and AD. The pattern of atrophy and its progression also aligned with existing structural connectivity in FHAD. In AD, our findings suggest that atrophy progression results from pathology propagation that occurred earlier, on a previously intact connectome. Moreover, a relationship was found between serotonin receptor spatial distribution and atrophy progression in AD.
The current study demonstrates that regions showing atrophy progression in FHAD and AD present with specific connectivity and cellular characteristics, uncovering some of the mechanisms involved in preclinical and clinical neurodegeneration.
... To extract cortical thickness measures, T1w data were processed using the Advanced Normalization Tools (ANTs) antsCorticalThickness.sh pipeline (Tustison et al., 2014) (https://github.com/ftdc-picsl/antsct-aging [v0.3.3-p01]). ...
Network segregation, which facilitates functional specialization of cognition in healthy brains, breaks down with neurodegeneration and therefore is a candidate biomarker of disease. However, limited evidence exists evaluating the role of network desegregation in behavioral variant frontotemporal degeneration (bvFTD), especially the breakdown of structural networks and the impact of this process on cognition. We collected neuropsychological tests of executive function (digit span backwards, letter fluency, category fluency) and a control task (confrontation naming), and diffusion MRI data in a sample of 95 bvFTD patients without evidence of primary progressive aphasia and 72 age-matched cognitively normal controls (CNC). We evaluated the hypotheses that bvFTD would have desegregation in structural networks compared to CNC and that, in bvFTD, executive dysfunction would relate to desegregation of the salience network. Probabilistic tractography maps were generated from diffusion MRI, and network segregation was defined as greater within-network connectivity than between-network connectivity. One-way ANCOVAs tested for group differences in network desegregation. Then linear regressions examined associations between network desegregation and neuropsychological test performance. Analyses controlled for age, sex, education, mean cortical atrophy, motion during diffusion MRI scan, imaging protocol, and disease duration. Compared to CNC, patients with bvFTD exhibited desegregation of the salience (p < .001) and global brain network (p = .006). In bvFTD, desegregation of salience network was associated with worse executive function (pcorrected = .039) but not confrontation naming. Results demonstrate associations between executive dysfunction and salience network desegregation in patients with bvFTD. Our findings indicate that brain network desegregation, reflecting reduced neural capacity for specialized processing, may contribute to the emergence of executive dysfunction in bvFTD.
... DBM maps were derived from each participant's T1-weighted MRI image at each time point using the Advanced Normalization Tools Longitudinal Cortical Thickness pipeline [see 'Deformation-based morphometry (DBM) processing' in supplementary material]. 27 Quality control (QC) was done by visual inspection of the resultant DBM maps. In total, 158 subjects with Alzheimer's disease (19%), 40 FHAD subjects (18%) and 67 HC (18%) were excluded mostly due to inaccurate grey matter segmentation (see Table 1 for the remaining number of subjects at each time point). ...
Alzheimer’s disease (AD) is associated with presymptomatic changes in brain morphometry and accumulation of abnormal tau and amyloid-beta pathology. Studying the development of brain changes prior to symptoms onset may lead to early diagnostic biomarkers and a better understanding of AD pathophysiology. AD pathology is thought to arise from a combination of protein accumulation and spreading via neural connections, but how these processes influence brain atrophy progression in the presymptomatic phases remains unclear. Individuals with a family history of AD (FHAD) have an elevated risk of AD, providing an opportunity to study the presymptomatic phase.
Here we used structural MRI from three databases (Alzheimer’s Disease Neuroimaging Initiative, Pre-symptomatic Evaluation of Experimental or Novel Treatments for Alzheimer Disease and Montreal Adult Lifespan Study) to map atrophy progression in FHAD and AD and assess the constraining effects of structural connectivity on atrophy progression. Cross-sectional and longitudinal data up to four years were used to perform atrophy progression analysis in FHAD and AD compared to controls. Positron emission tomography radiotracers were also used to quantify the distribution of abnormal tau and amyloid-beta protein isoforms at baseline. We first derived cortical atrophy progression maps using deformation-based morphometry from 153 FHAD, 156 AD, and 116 controls with similar age, education, and sex at baseline. We next examined the spatial relationship between atrophy progression and spatial patterns of tau aggregates and amyloid-beta plaques deposition, structural connectivity, and neurotransmitter receptor and transporter distributions.
Our results show that there were similar patterns of atrophy progression in FHAD and AD, notably in the cingulate, temporal, and parietal cortices, with more widespread and severe atrophy in AD. Both tau and amyloid-beta pathology tended to accumulate in regions that were structurally connected in FHAD and AD. The pattern of atrophy and its progression also aligned with existing structural connectivity in FHAD. In AD, our findings suggest that atrophy progression results from pathology propagation that occurred earlier, on a previously intact connectome. Moreover, a relationship was found between serotonin receptor spatial distribution and atrophy progression in AD.
The current study demonstrates that regions showing atrophy progression in FHAD and AD present with specific connectivity and cellular characteristics, uncovering some of the mechanisms involved in preclinical and clinical neurodegeneration.
... Cortical thickness was assessed with FreeSurfer, which calculates cortical thickness of a brain region using structural MRI data. 21,22 Using previously described methods, a 3-dimensional cortical surface model was created from Tl-weighted MRI with the automatic FreeSurfer "recon-all" pipeline. 21,23 Surface models of the outer limit of the grey matter (pial surface) and white matter were mapped onto the MRI. ...
Background
Non-Alzheimer's disease dementias, including frontotemporal dementia (FTD) can be difficult to characterize due to the predominance of distinct behavioral and neuropsychiatric symptoms. Widely used measurement tools lack structure and objectivity.
Objective
The purpose of this study was to use systematic direct observation of neuropsychiatric and behavioral symptoms, via the Neurobehavioral Rating Scale (NBRS), to characterize clusters of behavioral and neuropsychiatric symptoms in FTD and examine how selected symptom clusters correlate with structural neuroimaging.
Methods
We performed a factor analysis on the NBRS data from 172 patients with FTD and examined the neural correlates of the selected symptom clusters in a subsample of 67 patients.
Results
Six factors accounted for 56% of total variance across NBRS item scores: Apathy/Blunting, Agitation/Disinhibition, Cognitive/Language, Planning/Insight, Anxiety/Lability, and Psychosis. Symptom clusters showed significant associations with specific regions of cortical thinning: Agitation/Disinhibition with bilateral frontal regions, and Cognition/Language with the left bank of the superior temporal sulcus and supramarginal regions.
Conclusions
The selected symptom clusters associated with known regions of atrophy in FTD. The NBRS is an effective observational measure that may extend characterization and understanding of FTD.
... We non-rigidly registered CT images from paired 20-second breath-hold PET/CT images and the corresponding 300-second PET/CT images to get a deformation field by Advanced Normalization Tools (ANTs) [22]. Then we applied the deformation field to the 300-second PET acquisition to get the reference full-time PET image [23]. ...
Purpose
Respiratory motion during PET acquisition may produce lesion blurring. Ultra-fast 20-second breath-hold (U2BH) PET reduces respiratory motion artifacts, but the shortened scanning time increases statistical noise and may affect diagnostic quality. This study aims to denoise the U2BH PET images using a deep learning (DL)-based method.
Methods
The study was conducted on two datasets collected from five scanners where the first dataset included 1272 retrospectively collected full-time PET data while the second dataset contained 46 prospectively collected U2BH and the corresponding full-time PET/CT images. A robust and data-efficient DL method called mask vision transformer (Mask-ViT) was proposed which, after fine-tuned on a limited number of training data from a target scanner, was directly applied to unseen testing data from new scanners. The performance of Mask-ViT was compared with state-of-the-art DL methods including U-Net and C-Gan taking the full-time PET images as the reference. Statistical analysis on image quality metrics were carried out with Wilcoxon signed-rank test. For clinical evaluation, two readers scored image quality on a 5-point scale (5 = excellent) and provided a binary assessment for diagnostic quality evaluation.
Results
The U2BH PET images denoised by Mask-ViT showed statistically significant improvement over U-Net and C-Gan on image quality metrics (p < 0.05). For clinical evaluation, Mask-ViT exhibited a lesion detection accuracy of 91.3%, 90.4% and 91.7%, when it was evaluated on three different scanners.
Conclusion
Mask-ViT can effectively enhance the quality of the U2BH PET images in a data-efficient generalization setup. The denoised images meet clinical diagnostic requirements of lesion detectability.
... The normalization procedure employed different methods for the brain, and for brainstem and cord regions. For brain regions, the normalization process uses the python package "dipy" (https://dipy.org/documentation/1.5.0/documentation/) which is based on the ANTs (Advanced Normalization Tools) software [51,52]. Normalization of brainstem and spinal cord regions has been described previously [26,53] and involves mapping sections of the template to the image data for brainstem regions with distinct anatomical features. ...
Altered neural signaling in fibromyalgia syndrome (FM) was investigated with functional magnetic resonance imaging (fMRI). We employed a novel fMRI network analysis method, Structural and Physiological Modeling (SAPM), which provides more detailed information than previous methods. The study involved brain fMRI data from participants with FM (N = 22) and a control group (HC, N = 18), acquired during a noxious stimulation paradigm. The analyses were supported by fMRI data from the brainstem and spinal cord in FM and HC, brain fMRI data from participants with provoked vestibulodynia (PVD), and eye-tracking data from an fMRI study of FM. The results demonstrate differences in connectivity, and in blood oxygenation-level dependent (BOLD) responses, between FM and HC. In the FM group, BOLD signals underwent a large increase during the first 40 seconds of each fMRI run, prior to the application of any stimuli, compared to much smaller increases in HC. This indicates a heightened state of neural activity in FM that is sustained during fMRI runs, and dissipates between runs. The exaggerated initial rise was not observed in PVD. Autonomic functioning differed between groups. Pupil sizes were larger in FM than in HC, and the groups exhibited pupil dilation to the same levels during noxious stimulation. The initial BOLD increase varied in relation to state and trait anxiety scores. The results indicate that people with FM enter a heightened state of neural activity associated with anxiety and autonomic functioning, during every fMRI run, concurrent with increased pupil sizes, and heightened pain sensitivity. These findings may relate to the well-known hypervigilance and global hypersensitivity of FM participants.
... Thickness measurements were performed bilaterally across predefined brain regions. Lobe-specific measurements were calculated by combining cortical regions into standardized anatomical lobes, allowing for an analysis of thickness variations within broader brain structures (Tustison et al. 2014). Finally, the VolBrain2 pipeline automatically generated reports that included volumetric data, asymmetry ratios, and cortical thickness values for each segmented brain region (Manjón and Coupé, 2016). ...
Much brain imaging work has underscored the functional connections among the inferior frontal gyrus (IFG; articulation), supramarginal gyrus (SMG; letter-sound correspondence), superior temporal gyrus (STG; sound) and fusiform gyrus (FFG; print) during basic reading processes. This reading network supports and coordinates the complex processes that contribute to successful reading. In line with the Hebbian notion that ‘neurons that fire together, wire together’ we examined cortical thickness among these regions and the extent to which these regions showed structural relationships in average and impaired readers. Structural high resolution brain scans from 87 adult participants including average (N = 56; 51 right-handed; females = 29; mean age = 20.5; SD = 2.14) and impaired (N = 31; 27 right-handed; females = 24; mean age = 23.1; SD = 4.23) readers were collated. Cortical thickness measurements of the left and right IFG, SMG, STG, and FFG were extracted. Average readers had significantly greater cortical thickness in the right IFG and right SMG compared to impaired readers. Within each group, similarly strong relationships between the left and right structures were observed. Average readers had a significantly stronger connection between the left IFG-FFG compared to impaired readers (p = 0.012). In contrast, the impaired readers had a significantly stronger connection between the left STG-FFG compared to average readers (p = 0.027). In conclusion, the findings suggest that structural relationships within the reading network may contribute to variations in reading proficiency, with average readers exhibiting distinct patterns of cortical thickness and relationships compared to impaired readers. Further exploration of these structural differences could offer valuable insights into the neural mechanisms underlying reading abilities and disabilities.
... From each participant, we acquired a structural T1-weighted MPRAGE MRI using a Siemens Trio 3.0T scanner with an 8-channel phased-array head coil with the following parameters: repetition time (TR) = 1,620 ms, echo time (TE) = 3 ms, slice thickness = 1.0 mm, flip angle = 15°, matrix = 192 × 256, and inplane resolution = 0.9 mm × 0.9 mm. T1 MRI images were preprocessed and gray matter density was calculated for each of 119 regions (see "Label construction" below) using the Advanced Normalization Tools (ANTS) (Avants et al., 2008, Avants et al., 2011 CorticalThickness package (Tustison et al., 2014). which incorporates the highly accurate (Klein et al., 2010) Advanced Normalization Tools (ANTs). ...
Multimodal neuroimaging studies of corticobasal syndrome using volumetric MRI and DTI successfully discriminate between Alzheimer's disease and frontotemporal lobar degeneration but this evidence has typically included clinically heterogeneous patient cohorts and has rarely assessed the network structure of these distinct sources of pathology. Using structural MRI data, we identify areas in fronto-temporo-parietal cortex with reduced gray matter density in corticobasal syndrome relative to age matched controls. A support vector machine procedure demonstrates that gray matter density poorly discriminates between frontotemporal lobar degeneration and Alzheimer's disease pathology subgroups with low sensitivity and specificity. In contrast, a statistic of local network efficiency demonstrates excellent discriminatory power, with high sensitivity and specificity. Our results indicate that the underlying pathological sources of corticobasal syndrome can be classified more accurately using graph theoretical statistics of white matter microstructure in association cortex than by regional gray matter density alone. These results highlight the importance of a multimodal neuroimaging approach to diagnostic analyses of corticobasal syndrome and suggest that distinct sources of pathology mediate the circuitry of brain regions affected by corticobasal syndrome.
... Study-specific tissue priors were created using a multi-atlas segmentation procedure . Next, each subject's high-resolution structural image was processed using the ANTs Cortical Thickness Pipeline (Tustison et al., 2014). Following bias field correction (Tustison et al., 2010), each structural image was diffeomorphically registered to the study-specific PNC template using the top-performing SyN deformation (Klein et al., 2009). ...
Since initial reports regarding the impact of motion artifact on measures of functional connectivity, there has been a proliferation of confound regression methods to limit its impact. However, recent techniques have not been systematically evaluated using consistent outcome measures. Here, we provide a systematic evaluation of 12 commonly used confound regression methods in 193 young adults. Specifically, we compare methods according to three benchmarks, including the residual relationship between motion and connectivity, distance-dependent effects of motion on connectivity, and additional degrees of freedom lost in confound regression. Our results delineate two clear trade-offs among methods. First, methods that include global signal regression minimize the relationship between connectivity and motion, but unmask distance-dependent artifact. In contrast, censoring methods mitigate both motion artifact and distance-dependence, but use additional degrees of freedom. Taken together, these results emphasize the heterogeneous efficacy of proposed methods, and suggest that different confound regression strategies may be appropriate in the context of specific scientific goals.
... Next, T1w and T2w images were bias-field corrected using ANTs N4ITK [30]. For each subject, the T1w image was rigidly registered to the corresponding T2w image using ANTs registration [31]. The images were then skull-stripped and segmented using the dHCP anatomical pipeline [32]. ...
While the newborn’s brain is functionally organised early on—with similar resting state networks as those of adults present at birth—these networks continue to develop at different rates and in complex ways over time. While most in vivo functional imaging studies examine the spatial characteristics of resting state networks (RSNs), such as their size or the degree of connectivity, the temporal characteristics of these networks are less well characterised. We set out to examine the long-range temporal correlation (LRTC) of the fMRI blood oxygen level-dependent (BOLD) signal using the Hurst exponent in various RSNs in infants born very preterm shortly after birth (< 32 weeks gestational age; n = 64) and again later at term equivalent age (TEA) (n = 69). The Hurst exponent in grey matter and white matter was 0.69 at preterm age and increased significantly to 0.80 at TEA, with a greater difference between the tissues at TEA. The Hurst exponent in RSNs similarly was found to be approximately 0.68 for most networks at preterm age but increased significantly at different rates by TEA: 0.77 and 0.76 in the cerebellum and frontal networks respectively, and 0.84 and 0.83 in the motor and visual networks respectively. This pattern is partly consistent with findings from previous functional connectivity fMRI studies that the general growth and maturation of RSNs occur first and develop more quickly in sensory and motor networks, but later in associative networks like frontal ones. Importantly, this is the first time that this pattern has been shown in the LRTC of the fMRI BOLD signal itself, an area of study that may provide greater insight into functional brain development.
... To avoid this problem, we went through several steps. For each anatomical scan, we looked at the distribution of intensity values using ITK-SNAP [53] and then clipped them accordingly; then we rescaled those values using ANTs [54]; and then we performed denoising and bias correction using SPM12 and cat12. The surface reconstruction results from a fMRIprep anat only process were then carefully checked and improved where needed. ...
Previous studies demonstrated the existence of hand and tool areas in lateral and ventral occipitotemporal cortex (OTC), as well as an overlap between them. We reinvestigated this organization using 7T fMRI, benefiting from a higher signal-to-noise ratio than 3T. This enabled us to include a wider array of categories to achieve a more holistic perspective, and to omit certain spatial preprocessing steps. Despite these improvements, univariate analysis confirmed the existence of hand-tool overlap across OTC, which is striking given the omission of the spatial preprocessing steps that can influence overlap. There was significantly more overlap between hands and tools, compared to other overlap types in the left hemisphere of OTC. The overlap was also larger in the left lateral OTC as compared to the right lateral OTC. We found in all hand areas a differentiation between tools and other types of objects, although they still responded more to bodies than to tools. Regarding the tool areas, we observed a differentiation between hands and other categories such as faces and non-tool objects. Left hemisphere tool areas also differentiated between hands and bodies. When excluding the overlapping voxels from the hand and tool areas, they still showed a significant response to tools or hands (compared to objects or faces) respectively. Multi-voxel pattern analysis indicated that neural representations in the hand areas showed greater similarity between hands and tools than between hands and other objects. In the tool areas, the neural representations between tools and hands and between tools and other type of objects were all equally similar. To summarize, capitalizing on the benefits of 7T fMRI, we further substantiate the evidence in favor of hand-tool overlap in several regions of occipitotemporal cortex.
... It is generally accepted that there is a strong relationship between age and brain structures (Dima et al. 2022;Masouleh et al. 2019) and thus valid subcortical volume measures should be able to predict age accurately. Age prediction accuracy has been used as a measure of neurobiological validity and compared across software packages or analysis strategies in previous studies (Sebenius et al. 2023;Tustison et al. 2014). In this manner, we aimed to provide preliminary evidence suggesting which software package might offer greater accuracy. ...
There is still little research on the consistency among the subcortical volume estimates of different software packages. It is also unclear whether there are age‐related differences in the inter‐software consistency. The current study aimed to examine the consistency of three commonly used automated software packages and the effect of age on inter‐software consistency. We analyzed T1‐weighted structural images from two public datasets, in which the subjects were divided into four age groups ranging from childhood and adolescence to late adulthood. We chose three mainstream automated software packages including FreeSurfer, CAT, and FSL, to estimate the volumes of seven subcortical structures, including thalamus, caudate, putamen, pallidum, hippocampus, amygdala, and accumbens. We used the intraclass correlation coefficient (ICC) and Pearson correlation coefficient (PCC) to quantify inter‐software consistency and compared the consistency measures among the age groups. As a measure of validity, we additionally evaluated the predictive power of each software package's estimates for predicting age. The results showed good inter‐software consistency in the thalamus, caudate, putamen, and hippocampus, moderate consistency in the pallidum, and poor consistency in the amygdala and accumbens. Significant differences in the inter‐software consistency were not observed among the age groups in most cases. FreeSurfer exhibited higher age prediction accuracy than CAT and FSL. The current study showed that the inter‐software consistency on the subcortical volume estimation varies with structures but generally not with age groups, which has important implications for the interpretation and reproducibility of neuroimaging findings.
... Details of the study rationale, protocol, and baseline characteristics have been published elsewhere. 3,27 We recruited consecutive patients ≥18 years old who had experienced an acute stroke of any stroke severity, with symptom onset within the last 5 days and no pre- 29 The difference images were calculated by subtracting the intensity-normalized images at baseline from the registered 6-month follow-up images. All images were evaluated in a standardized reading setup ( Figure 1A, Video S1, and Figure S1). ...
INTRODUCTION
While incident ischemic lesions (IILs) are not unusual on follow‐up magnetic resonance imaging (MRI) following stroke, their risk factors and prognostic significance remain unknown.
METHODS
In a prospective multicenter study of 503 acute stroke patients, we assessed IILs on registered MRI images at baseline and 6 months, analyzing risk factors and clinical outcomes across 36 months.
RESULTS
At 6 months, 78 patients (15.5%) had IILs, mostly diffusion‐weighted imaging‐positive (72%) and clinically covert (91%). Older age and small vessel disease (SVD) lesions were baseline risk factors for IILs. IILs were associated with worse cognitive (beta for global cognition: −0.31, 95% confidence interval [CI]: −0.48 to −0.14) and functional outcomes (beta for modified Rankin scale [mRS]: 0.36, 95% CI: 0.14 to 0.58), and higher recurrent stroke risk (hazard ratio: 3.81, 95% CI: 1.35 to 10.69). IILs partially explained the relationship between SVD and poor cognition.
DISCUSSION
IILs are common and are associated with worse cognitive and functional outcomes and stroke recurrence risk. Assessing IILs following stroke might aid prognostication.
Highlights
Incident ischemic lesions (IILs) were assessed with registered baseline and 6‐month magnetic resonance imaging (MRI) scans in a stroke cohort.
IILs 6 months after stroke are present in one‐sixth of patients and are mostly clinically silent.
Small vessel disease burden is the main baseline risk factor for IILs.
IILs are associated with cognitive and functional impairment and stroke recurrence.
Assessing IILs by follow‐up MRI aids long‐term prognostication for stroke patients.
... Where t s is the morphometric measurement of a parcel for session s and t is the mean of t across sessions 55,79 . Thus, we defined variability as the mean absolute percent difference between each individual and the mean across sessions. ...
Pregnancy is a period of profound hormonal and physiological changes experienced by millions of women annually, yet the neural changes unfolding in the maternal brain throughout gestation are not well studied in humans. Leveraging precision imaging, we mapped neuroanatomical changes in an individual from preconception through 2 years postpartum. Pronounced decreases in gray matter volume and cortical thickness were evident across the brain, standing in contrast to increases in white matter microstructural integrity, ventricle volume and cerebrospinal fluid, with few regions untouched by the transition to motherhood. This dataset serves as a comprehensive map of the human brain across gestation, providing an open-access resource for the brain imaging community to further explore and understand the maternal brain.
... The s-MRI used as inputs for the contrastive machine learning remain in native space, i.e., they are not postprocessed using any specific brain templates and are normalized between 0 and 1, and then resampled to a resolution of 64 pixel 64 pixel 64 pixel before being fed into the CVAE for feature extraction. While in Section 3.3, the s-MRIs are post-processed into features of different cortical regions: for children (0−2)-year-old, the children's s-MRIs are processed by the Infantfreesurfer [21] tools, based on the per-month strategy (i.e., (0−2)-year-old, range from 0 to 24 months, is divided into 24 intervals, with each interval representing the length of a month, and children's s-MRIs within each interval are processed separately); for children (2−5)-year-old, the children's s-MRIs are first projected onto the age-specific brain templates to create the prior masks using ANTs [22] , and then further processed by Freesurfer [23] tools; the Freesurfer tools automatically parcellate the cortex and assign a neuroanatomical label to each location of the cortex (i.e., gray matter) based on probabilistic information, and the cortex is divided into 34 regions in this study based on Desikan-Killiany atlas [24] ; surface area of each cortical region is measured as the feature indicating its uniqueness. All data used in this study are approved by local Internal Review Boards and carried out in accordance with ethics guidelines and regulations, and informed consents are obtained from all participants, or parent/legal guardians if participants are under 18-year-old. ...
Autism Spectrum Disorder (ASD) is a highly disabling mental disease that brings significant impairments of social interaction ability to the patients, making early screening and intervention of ASD critical. With the development of the machine learning and neuroimaging technology, extensive research has been conducted on machine classification of ASD based on structural Magnetic Resonance Imaging (s-MRI). However, most studies involve with datasets where participants’ age are above 5 and lack interpretability. In this paper, we propose a machine learning method for ASD classification in children with age range from 0.92 to 4.83 years, based on s-MRI features extracted using Contrastive Variational AutoEncoder (CVAE). 78 s-MRIs, collected from Shenzhen Children’s Hospital, are used for training CVAE, which consists of both ASD-specific feature channel and common-shared feature channel. The ASD participants represented by ASD-specific features can be easily discriminated from Typical Control (TC) participants represented by the common-shared features. In case of degraded predictive accuracy when data size is extremely small, a transfer learning strategy is proposed here as a potential solution. Finally, we conduct neuroanatomical interpretation based on the correlation between s-MRI features extracted from CVAE and surface area of different cortical regions, which discloses potential biomarkers that could help target treatments of ASD in the future.
... FSL's FAST was used for tissue segmentation of cerebrospinal fluid (CSF), white matter (WM), and gray matter (GM). Using antsRegistration (26), the T1w reference image was nonlinearly registered to the ICBM 152 Nonlinear Asymmetrical template version 2009c template. ...
Introduction
Frontotemporal lobar degeneration (FTLD) is associated with FTLD due to tau (FTLD-tau) or TDP (FTLD-TDP) inclusions found at autopsy. Arterial Spin Labeling (ASL) MRI is often acquired in the same session as a structural T1-weighted image (T1w), enabling detection of regional changes in cerebral blood flow (CBF). We hypothesize that ASL-T1w registration with more degrees of freedom using boundary-based registration (BBR) will better align ASL and T1w images and show increased sensitivity to regional hypoperfusion differences compared to manual registration in patient participants. We hypothesize that hypoperfusion will be associated with a clinical measure of disease severity, the FTLD-modified clinical dementia rating scale sum-of-boxes (FTLD-CDR).
Materials and methods
Patients with sporadic likely FTLD-tau (sFTLD-tau; N = 21), with sporadic likely FTLD-TDP (sFTLD-TDP; N = 14), and controls (N = 50) were recruited from the Connectomic Imaging in Familial and Sporadic Frontotemporal Degeneration project (FTDHCP). Pearson’s Correlation Coefficients (CC) were calculated on cortical vertex-wise CBF between each participant for each of 3 registration methods: (1) manual registration, (2) BBR initialized with manual registration (manual+BBR), (3) and BBR initialized using FLIRT (FLIRT+BBR). Mean CBF was calculated in the same regions of interest (ROIs) for each registration method after image alignment. Paired t-tests of CC values for each registration method were performed to compare alignment. Mean CBF in each ROI was compared between groups using t-tests. Differences were considered significant at p < 0.05 (Bonferroni-corrected). We performed linear regression to relate FTLD-CDR to mean CBF in patients with sFTLD-tau and sFTLD-TDP, separately (p < 0.05, uncorrected).
Results
All registration methods demonstrated significant hypoperfusion in frontal and temporal regions in each patient group relative to controls. All registration methods detected hypoperfusion in the left insular cortex, middle temporal gyrus, and temporal pole in sFTLD-TDP relative to sFTLD-tau. FTLD-CDR had an inverse association with CBF in right temporal and orbitofrontal ROIs in sFTLD-TDP. Manual+BBR performed similarly to FLIRT+BBR.
Discussion
ASL is sensitive to distinct regions of hypoperfusion in patient participants relative to controls, and in patients with sFTLD-TDP relative to sFTLD-tau, and decreasing perfusion is associated with increasing disease severity, at least in sFTLD-TDP. BBR can register ASL-T1w images adequately for controls and patients.
Regional brain atrophy estimated from structural magnetic resonance imaging (MRI) is a widely used measure of neurodegeneration in Alzheimer’s disease (AD), Frontotemporal Lobar Degeneration (FTLD), and other dementias. Yet, traditional MRI-derived morphometric estimates are susceptible to measurement errors, posing a challenge for reliably detecting longitudinal atrophy, particularly over short intervals. Here, we examined the utility of multiple MRI scans acquired in rapid succession (i.e., cluster scanning ) for detecting longitudinal cortical atrophy over 3- and 6-month intervals within individual patients. Four individuals with mild cognitive impairment or mild dementia likely due to AD or FTLD participated in this study. At baseline, 3 months, and 6 months, structural MRI data were collected on a 3 Tesla scanner using a fast 1.2-mm T1-weighted multi-echo magnetization-prepared rapid gradient echo (MEMPRAGE) sequence (acquisition time = 2’23’’). At each timepoint, participants underwent up to 32 MEMPRAGE scans acquired in four separate sessions over two days. Using linear mixed-effects models, phenotypically vulnerable cortical (“core atrophy”) regions exhibited statistically significant longitudinal atrophy in all participants (i.e., decreased cortical thickness) by 3 months and further demonstrated preferential vulnerability compared to control regions in three of the participants over at least one of the 3-month intervals. These findings provide proof-of-concept evidence that pooling multiple morphometric estimates derived from cluster scanning can detect longitudinal cortical atrophy over short intervals in individual patients with neurodegenerative dementias.
Magnetic resonance images (MRI) of the brain exhibit high dimensionality that pose significant challenges for computational analysis. While models proposed for brain MRIs analyses yield encouraging results, the high complexity of neuroimaging data hinders generalizability and clinical application. We introduce DUNE, a neuroimaging-oriented encoder designed to extract deep-features from multisequence brain MRIs, thereby enabling their processing by basic machine learning algorithms. A UNet-based autoencoder was trained using 3,814 selected scans of morphologically normal (healthy volunteers) or abnormal (glioma patients) brains, to generate comprehensive low-dimensional representations of the full-sized images. To evaluate their quality, these embeddings were utilized to train machine learning models to predict a wide range of clinical variables. Embeddings were extracted for cohorts used for the model development (n=21,102 individuals), along with 3 additional independent cohorts (Alzheimer’s disease, schizophrenia and glioma cohorts, n=1,322 individuals), to evaluate the model’s generalization capabilities. The embeddings extracted from healthy volunteers’ scans could predict a broad spectrum of clinical parameters, including volumetry metrics, cardiovascular disease (AUROC=0.80) and alcohol consumption (AUROC=0.99), and more nuanced parameters such as the Alzheimer’s predisposing APOE4 allele (AUROC=0.67). Embeddings derived from the validation cohorts successfully predicted the diagnoses of Alzheimer’s dementia (AUROC=0.92) and schizophrenia (AUROC=0.64). Embeddings extracted from glioma scans successfully predicted survival (C-index=0.608) and IDH molecular status (AUROC=0.92), matching the performances of previous task-oriented models. DUNE efficiently represents clinically relevant patterns from full-size brain MRI scans across several disease areas, opening ways for innovative clinical applications in neurology.
One Sentence Summary
We propose a brain MRI-specialized encoder, which extracts versatile low-dimension embeddings from full-size scans.
Background
A novel neuroimaging signature of regional cortical thickness on brain MRI recently showed high potential for Alzheimer's disease and related dementias (ADRD) risk stratification in the community. How these findings translate to other populations, remains undetermined.
Objective
We aimed to replicate this novel ADRD neuroimaging marker in the population-based Rotterdam Study.
Methods
We included all participants from the population-based Rotterdam Study with brain-MRI between 2005–2016, and derived the signature using FreeSurfer. We computed hazard ratios and C-statistics for 10-year dementia risk, and betas for cross-sectional associations with cognition, comparing the novel signature to hippocampal volume, mean cortical thickness, and another cortical thickness signature (Dickerson's).
Results
Of 3249 participants (mean age 71.3 ± 8.0 years), 294 developed dementia (74.8% clinical AD) during a mean follow-up of 8.1 years. The novel ADRD signature had similar magnitude of associations as Dickerson's signature and cortical thickness for AD dementia (HR per 1-SD increase 0.87;0.78–0.96), but performed worse than all markers for all-cause dementia. Of the four neuroimaging markers, hippocampal volume showed the strongest associations with both risk of all-cause dementia and clinical AD dementia. The ADRD had the weakest association with general cognitive function (β per 1-SD increase 0.04;0.02–0.06), and executive function (β per 1-SD increase 0.02;0.00–0.04), followed by cortical thickness and Dickerson's, and hippocampal volume showed the strongest associations.
Conclusions
In this community-based study, the novel cortical thickness signature did not outperform hippocampal volume for dementia risk stratification. The importance of replication studies underlines the value of the current study. Replicating research findings is essential to establish robust biomarkers for dementia risk prediction.
The dynamic integration of the lateralized and specialized capacities of the two cerebral hemispheres constitutes a hallmark feature of human brain function. This inter-hemispheric exchange of information is thought to critically depend upon the corpus callosum. Classical anatomical descriptions of callosal organization outline a topographic gradient from front to back, such that specific transcallosal fibers support distinct aspects of integrated brain function. Here we present a challenge to this conventional model. Using neuroimaging data obtained from a new cohort of adult callosotomy patients, we leverage modern network neuroscience techniques to show - for the first time - that full inter-hemispheric integration can be achieved via a small proportion of posterior callosal fibers. Partial callosotomy patients with spared callosal fibers retained widespread patterns of inter-hemispheric functional connectivity and showed no signs of behavioral disconnection syndromes, even with only 1 cm of the splenium intact. Conversely, only complete corpus callosotomy patients demonstrated sweeping disruptions of inter-hemispheric network architectures, aligning with disconnection syndromes long-thought to reflect diminished information propagation and communication across the brain. These findings motivate a novel mechanistic understanding of synchronized inter-hemispheric neural activity for large-scale human brain function and behavior.
In this paper, we introduce holiAtlas, a holistic, multimodal and high-resolution human brain atlas. This atlas covers different levels of details of the human brain anatomy, from the organ to the substructure level, using a new dense labelled protocol generated from the fusion of multiple local protocols at different scales. This atlas has been constructed averaging images and segmentations of 75 healthy subjects from the Human Connectome Project database. Specifically, MR images of T1, T2 and WMn (White Matter nulled) contrasts at 0.125 resolution that were nonlinearly registered and averaged using symmetric group-wise normalisation to construct the atlas. At the finest level, the holiAtlas protocol has 350 different labels derived from 10 different delineation protocols. These labels were grouped at different scales to provide a holistic view of the brain at different levels in a coherent and consistent manner. This multiscale and multimodal atlas can be used for the development of new ultra-high resolution segmentation methods that can potentially leverage the early detection of neurological disorders.
Substantia nigra (SN) and locus coeruleus (LC) are two catecholaminergic, neuromelanin-rich nuclei that are affected in Parkinson’s Disease and may show neuroimaging abnormalities before the onset of motor manifestations. The simultaneous, multimodal investigation of their microstructural abnormalities may provide useful insights on the spatial diffusion and tissue characateristics of neurodegeneration, and this may in turn help develop markers for disease-modifying clinical trials. Therefore, through neuromelanin-sensitive and diffusion MRI, we aimed to investigate microstructural abnormalities in those nuclei in isolated REM sleep behaviour disorder (iRBD) and Parkinson’s Disease (PD).
Fourteen participants with polysonomnography-confirmed iRBD, 18 with PD and 18 healthy controls were scanned with structural, neuromelanin-sensitive and neurite orientation dispersion and density imaging (NODDI) MRI. iRBD participants also underwent dopamine transporter imaging. SN neuromelanin and NODDI diffusion parameters and LC neuromelanin signals were extracted. Motor and global cognitive assessments were also collected.
iRBD and PD participants showed significantly reduced neuromelanin contrast in the LC middle section compared to healthy controls. PD also showed significantly reduced caudal LC and posterior SN neuromelanin signal. No differences in SN NODDI parameters were detected between iRBD and healthy controls. Five iRBD participants showed reduced striatal dopamine transporter. In the combined disease groups (iRBD and PD), significant associations were shown between SN neuromelanin signal and neurite density index (r=-0.610, corr-p=0.001), and between SN neurite density index and free water fraction (r=0.417, corr-p=0.042). In the same group, motor scores were negatively associated with nigral neuromelanin signal (r=-0.404, corr-p=0.044) and free water fraction (r=0.486, corr-p=0.018).
In conclusion, iRBD participants showed significant neuromelanin loss in the LC, with a minority showing initial nigrostriatal dopaminergic abnormalities. Across the entire iRBD-PD spectrum, the association between SN neuromelanin signal loss, diffusion parameters and motor scores has the potential to capture different yet related aspects of SN degeneration.
Low-dose (LD) PET imaging would lead to reduced image quality and diagnostic efficacy. We propose a deep learning (DL) method to reduce radiotracer dosage for PET studies while maintaining diagnostic quality.
This retrospective study was performed on 456 participants respectively scanned by three different PET scanners with two different tracers. A DL method called spatially aware noise reduction network (SANR) was proposed to recover 3D full-dose (FD) PET volumes from LD data. The performance of SANR was compared with a 2D DL method taking regular FD PET volumes as the reference. Wilcoxon signed-rank test was conducted to compare the image quality metrics across different DL denoising methods. For clinical evaluation, two nuclear medicine physicians examined the recovered FD PET volumes using a 5-point grading scheme (5 = excellent) and gave a binary decision (negative or positive) for diagnostic quality assessment.
Statistically significant differences (p < 0.05) were found in terms of image quality metrics when SANR was compared with the 2D DL method. For clinical evaluation, SANR achieved a lesion detection accuracy of 95.3% (95% CI: 90.1%, 100%), while the reference full-dose PET volumes obtained a lesion detection accuracy of 98.4% (95% CI: 95.4%, 100%). In Alzheimer’s disease diagnosis, both the reference FD PET volumes and the FD PET volumes recovered by SANR exhibited the same accuracy.
Compared with reference FD PET, LD PET denoised by the proposed approach significantly reduced radiotracer dosage and showed noninferior diagnostic performance in brain lesion detection and Alzheimer’s disease diagnosis.
Question The current trend in PET imaging is to reduce injected dosage, which leads to low-quality PET images and reduces diagnostic efficacy.
Findings The proposed deep learning method could recover diagnostic quality PET images from data acquired with a markedly reduced radiotracer dosage.
Clinical relevance The proposed method would enhance the utility of PET scanning at lower radiotracer dosage and inform future workflows for brain lesion detection and Alzheimer’s disease diagnosis, especially for those patients who need multiple examinations.
Polypathology is a major driver of heterogeneity in clinical presentation and extent of neurodegeneration (N) in patients with Alzheimer Disease (AD). Beyond amyloid (A) and tau (T) pathologies, over half of patients with AD have concomitant pathology such as α-synuclein (S) in mixed AD with Lewy Body Disease (LBD). Patients with Mixed Etiology Dementia (MED) such as AD+LBD have faster progression and potentially differential responses to targeted treatments, though the diagnosis of AD+LBD can be challenging given overlapping clinical and imaging features. Development and validation of improved in vivo biomarkers are required to study relationships between N and S and identify imaging patterns reflecting mixed AD+LBD pathologies.
We hypothesize that individual proteinopathies, such as T and S, are associated with commensurate levels of N. Thus, we assessed biomarkers of A, T, N and S with positron emission tomography (PET), magnetic resonance imaging (MRI) and cerebrospinal fluid (CSF) seeding amplification assay (SAA) data to determine molecular presentations of mixed A+S+ vs. A+S– cognitively impaired patients from the Alzheimer’s Disease Neuroimaging Initiative (ADNI).
We found A+S+ patients had parieto-occipital 18F-Fluorodeoxyglucose hypometabolism (a measure of N) disproportionate to the degree of regional atrophy or T burden, highlighting worse hypometabolism associated with S+ SAA. Following up on this hypometabolic mismatch with CSF metabolite and proteome analyses, we found that A+S+ patients exhibited lower CSF levels of dopamine metabolites and synaptic markers like neuronal pentraxin-2 (NPTX2), suggesting that altered neurotransmission and neuron integrity contribute to this dissociation between metabolic PET and MRI. Potential confounders exist when studying relations between N, AD and LBD pathologies, including neuroinflammation and other non-Alzheimer pathologies in MED, though our findings imply posterior hypometabolic mismatch is related more to S than vascular or TDP-43 pathology.
A+S+ patients had posterior mismatch with excessive 18F-Fluorodeoxyglucose hypometabolism relative to atrophy or T load, possibly reflecting impaired neuron integrity. Further research must disentangle the impact of multiple proteinopathies and clinicopathologic factors on hypometabolism and atrophy. Cumulatively, patients with mixed AD+LBD etiologies harbor a unique metabolic PET mismatch signature.
One of the neurobiological correlates of alcohol use disorder (AUD) is the disruption of striatal dopaminergic function. Although regional differences in dopamine (DA) tone/function have been well studied, interregional relationships (represented as inter‐subject covariance) have not been investigated and may offer a novel avenue for understanding DA tone. Positron emission tomography (PET) data with [¹¹C]raclopride in 22 social drinking controls and 17 AUD participants were used to generate group‐level striatal covariance (partial Pearson correlation) networks, which were compared edgewise as well as on global network metrics and community structure. An exploratory analysis examined the impact of tobacco cigarette use status. Striatal covariance was validated in an independent publicly available [¹⁸F]fallypride PET sample of healthy volunteers. Striatal covariance of control participants from both data sets showed a clear bipartition of the network into two distinct communities, one in the anterior and another in the posterior striatum. This organization was disrupted in the AUD participants' network, which showed significantly lower network metrics compared with the control participants' network. Stratification by cigarette use suggests differential consequences on group covariance networks. This work demonstrates that network neuroscience can quantify group differences in striatal DA and that its interregional interactions offer new insight into the consequences of AUD.
Background
Brain metastasis invasion pattern (BMIP) is an emerging biomarker associated with recurrence-free and overall survival in patients, and differential response to therapy in preclinical models. Currently, BMIP can only be determined from the histopathological examination of surgical specimens, precluding its use as a biomarker prior to therapy initiation. The aim of this study was to investigate the potential of machine learning (ML) approaches to develop a non-invasive magnetic resonance imaging (MRI)-based biomarker for BMIP determination.
Methods
From an initial cohort of 329 patients, a subset of 132 patients met inclusion criteria for this retrospective study. We evaluated the ability of an expert neuroradiologist to reliably predict BMIP. Thereafter, the dataset was randomly divided into training/validation (80% of cases) and test subsets (20% of cases). Ground truth for BMIP was histopathologic evaluation of resected specimens. Following MRI sequence co-registration, advanced feature extraction techniques deriving hand-crafted radiomic features with traditional ML classifiers and convolution-based deep learning (CDL) models were trained and evaluated. Different ML approaches were used individually or using ensembling techniques to determine the model with the best performance for BMIP prediction.
Results
Expert evaluation of brain MRI scans could not reliably predict BMIP, with an accuracy of 44-59% depending on the semantic feature used. Among the different ML and CDL models evaluated, the best performing model achieved an accuracy of 85% and an F1-score of 90%.
Conclusion
ML approaches can effectively predict BMIP, representing a non-invasive MRI-based approach to guide management of patients with brain metastases.
Aim
Methamphetamine use disorders (MUDs) cause widespread disruptions in metabolomic and immunologic processes, highlighting the need for new therapeutic approaches. The purpose of this study was to find molecular and neuroimaging biomarkers for methamphetamine addiction.
Methods
In this study, we recruited 231 patients with MUD at varying stages of withdrawal and 40 healthy controls to quantify the blood levels of 52 molecules using enzyme‐linked immunosorbent assay.
Results
The overall molecular disruption caused by methamphetamine was inversely related to withdrawal time ( P = 0.0008), with partial recovery observed after 1 year of follow‐up ( P = 2.20 × 10 ⁻⁵ ). Molecules related to stress, immune activation, oxidative products, and cardiac injury were significantly elevated in all MUD groups, while antioxidation enzymes were downregulated. Additionally, the blood level of brain‐derived neurotrophic factor was significantly correlated with gray matter volumes in nine brain regions (fusiform gyrus, orbitofrontal cortex, temporal pole, caudate, cerebellum crus, and vermis, adjusted P < 0.05) among patients with MUD.
Conclusion
These findings suggest that patients with MUD exhibit elevated levels of immune response, stress, and oxidative stress, which are associated with brain structural abnormalities.
Importance
Prior studies associate late-life community disadvantage with worse brain health. It is relatively unknown if childhood community disadvantage associates with late-life brain health.
Objective
To test associations between childhood residence in an economically disadvantaged community, individual income and education, and late-life cortical brain volumes and white matter integrity.
Design, Setting, and Participants
This cohort study was conducted in the ongoing harmonized cohorts KHANDLE (Kaiser Healthy Aging and Diverse Life Experiences Study; initiated 2017) and STAR (Study of Healthy Aging in African Americans; initiated 2018) using all available data collected out of a regional integrated health care delivery network in California between cohort initiation and analysis initiation in June 2023. Eligible participants were Kaiser Permanente Northern California member ages 65 years or older. Data were analyzed between June and November 2023.
Exposure
Residence at birth was geocoded and linked to historical Area Deprivation Indices (ADI). ADI is a nationally ranked percentile; community disadvantage was defined as ADI of 80 or higher.
Main Outcomes and Measures
Regional brain volumes and white matter integrity measures were derived from a random subset of participants who underwent 3T magnetic resonance imaging. Models adjusted for race and ethnicity, sex, and parental education.
Results
Of a total 2161 individuals in the combined cohort, 443 individuals were eligible for imaging (mean [SD] age, 76.3 [6.5] years; 253 female [57.1%]; 56 Asian [12.6%], 212 Black [47.9%], 67 Latino [15.1%], 109 White [24.6%]). Imaging participants had a mean (SD) 15.0 (2.5) years of education, and 183 (41.3%) earned 99 999 annually. Fifty-four participants (12.2%) resided in a disadvantaged childhood community. Childhood community disadvantage was associated with smaller gray matter volumes overall (−0.39 cm ³ ; 95% CI, −0.65 to −0.10 cm ³ ) and in the cerebellum (−0.39 cm ³ ; 95% CI, −0.66 to −0.09 cm ³ ), hippocampus (−0.37 cm ³ ; 95% CI, −0.68 to −0.04 cm ³ ), and parietal cortex (−0.25 cm ³ ; 95% CI, −0.46 to −0.04 cm ³ ) and larger mean lateral ventricle (0.44 cm ³ ; 95% CI, 0.12 to 0.74 cm ³ ), third ventricle (0.28 cm ³ ; 95% CI, 0.03 to 0.55 cm ³ ), and white matter hyperintensity volume (0.31 cm ³ ; 95% CI, 0.06 to 0.56 cm ³ ). Educational attainment and late-life income did not mediate these associations.
Conclusions and Relevance
In this cohort study of racially and ethnically diverse health plan members, childhood community disadvantage was associated with worse late-life brain health independent of individual socioeconomic status. Future work should explore alternative pathways (eg, cardiovascular health) that may explain observed associations.
Evolutionarily relevant networks have been previously described in several mammalian species using time-averaged analyses of fMRI time-series. However, fMRI network activity is highly dynamic and continually evolves over timescales of seconds. Whether the dynamic organization of resting-state fMRI network activity is conserved across mammalian species remains unclear. Using frame-wise clustering of fMRI time-series, we find that intrinsic fMRI network dynamics in awake male macaques and humans is characterized by recurrent transitions between a set of 4 dominant, neuroanatomically homologous fMRI coactivation modes (C-modes), three of which are also plausibly represented in the male rodent brain. Importantly, in all species C-modes exhibit species-invariant dynamic features, including preferred occurrence at specific phases of fMRI global signal fluctuations, and a state transition structure compatible with infraslow coupled oscillator dynamics. Moreover, dominant C-mode occurrence reconstitutes the static organization of the fMRI connectome in all species, and is predictive of ranking of corresponding fMRI connectivity gradients. These results reveal a set of species-invariant principles underlying the dynamic organization of fMRI networks in mammalian species, and offer novel opportunities to relate fMRI network findings across the phylogenetic tree.
Proper quality control (QC) is time consuming when working with large-scale medical imaging datasets, yet necessary, as poor-quality data can lead to erroneous conclusions or poorly trained machine learning models. Most efforts to reduce data QC time rely on outlier detection, which cannot capture every instance of algorithm failure. Thus, there is a need to visually inspect every output of data processing pipelines in a scalable manner. We design a QC pipeline that allows for low time cost and effort across a team setting for a large database of diffusion weighted and structural magnetic resonance images. Our proposed method satisfies the following design criteria: 1.) a consistent way to perform and manage quality control across a team of researchers, 2.) quick visualization of preprocessed data that minimizes the effort and time spent on the QC process without compromising the condition or caliber of the QC, and 3.) a way to aggregate QC results across pipelines and datasets that can be easily shared. In addition to meeting these design criteria, we also provide information on what a successful output should be and common occurrences of algorithm failures for various processing pipelines. Our method reduces the time spent on QC by a factor of over 20 when compared to naively opening outputs in an image viewer and demonstrate how it can facilitate aggregation and sharing of QC results within a team. While researchers must spend time on robust visual QC of data, there are mechanisms by which the process can be streamlined and efficient.
Damage to the primary visual cortex or its afferent white-matter tracts results in loss of vision in the contralateral visual field that can present as homonymous visual field deficits. Evidence suggests that visual training in the blind field can partially reverse blindness at trained locations. However, the efficacy of visual training is highly variable across participants, and the reasons for this are poorly understood. It is likely that variance in residual neural circuitry following the insult may underlie the variation among patients. Many stroke survivors with visual field deficits retain residual visual processing in their blind field despite a lack of awareness. Previous research indicates that intact structural and functional connections between the dorsal lateral geniculate nucleus and the human extrastriate visual motion-processing area hMT+ is necessary for blindsight to occur. We therefore hypothesised that changes in this white matter pathway may underlie improvements resulting from motion discrimination training.
Eighteen stroke survivors with long-standing, unilateral, homonymous field defects from retro-geniculate brain lesions completed six months of visual training at home. This involved performing daily sessions of a motion discrimination task, at two non-overlapping locations in the blind field, at least five days per week. Motion discrimination and integration thresholds, Humphrey perimetry, and structural and diffusion-weighted MRI were collected pre- and post-training.
Changes in fractional anisotropy were analysed in visual tracts connecting the ipsilesional dorsal lateral geniculate nucleus and hMT+, and the ipsilesional dorsal lateral geniculate nucleus and primary visual cortex. The [non-visual] tract connecting the ventral posterior lateral nucleus of the thalamus and the primary somatosensory cortex was analysed as a control. Changes in white matter integrity were correlated with improvements in motion discrimination and Humphrey perimetry. We found that the magnitude of behavioural improvement was not directly related to changes in fractional anisotropy in the pathway between the dorsal lateral geniculate nucleus and hMT+ or dorsal lateral geniculate nucleus and primary visual cortex. Baseline fractional anisotropy in either tract also failed to predict improvements in training. However, an exploratory analysis showed a significant increase in fractional anisotropy in the distal part of the tract connecting the dorsal lateral geniculate nucleus and hMT+, suggesting that six months of visual training in chronic, retro-geniculate strokes may enhance white matter microstructural integrity of residual geniculo-extrastriate pathways.
Surface-based analysis of the cerebral cortex is ubiquitous in human neuroimaging with MRI. It is crucial for cortical registration, parcellation, and thickness estimation. Traditionally, these analyses require high-resolution, isotropic scans with good gray-white matter contrast, typically a 1mm T1-weighted scan. This excludes most clinical MRI scans, which are often anisotropic and lack the necessary T1 contrast. To enable large-scale neuroimaging studies using vast clinical data, we introduce recon-all-clinical, a novel method for cortical reconstruction, registration, parcellation, and thickness estimation in brain MRI scans of any resolution and contrast. Our approach employs a hybrid analysis method that combines a convolutional neural network (CNN) trained with domain randomization to predict signed distance functions (SDFs) and classical geometry processing for accurate surface placement while maintaining topological and geometric constraints. The method does not require retraining for different acquisitions, thus simplifying the analysis of heterogeneous clinical datasets. We tested recon-all-clinical on multiple datasets, including over 19,000 clinical scans. The method consistently produced precise cortical reconstructions and high parcellation accuracy across varied MRI contrasts and resolutions. Cortical thickness estimates are precise enough to capture aging effects independently of MRI contrast, although accuracy varies with slice thickness. Our method is publicly available at https://surfer.nmr.mgh.harvard.edu/fswiki/recon-all-clinical, enabling researchers to perform detailed cortical analysis on the huge amounts of already existing clinical MRI scans. This advancement may be particularly valuable for studying rare diseases and underrepresented populations where research-grade MRI data is scarce.
By honest I don't mean that you only tell what's true. But you make clear the entire situation. You make clear all the information that is required for somebody else who is intelligent to make up their mind.
Richard Feynman
The neuroscience community significantly benefits from the proliferation of imaging-related analysis software packages. Established packages such as SPM (Ashburner, 2012), the FMRIB Software Library (FSL) (Jenkinson et al., 2012), Freesurfer (Fischl, 2012), Slicer (Fedorov et al., 2012), and the AFNI toolkit (Cox, 2012) aid neuroimaging researchers around the world in performing complex analyses as part of ongoing neuroscience research. In conjunction with distributing robust software tools, neuroimaging packages also continue to incorporate algorithmic innovation for improvement in analysis tools.
As fellow scientists who actively participate in neuroscience research through our contributions to the Insight Toolkit1 (e.g., Johnson et al., 2007; Ibanez et al., 2009; Tustison and Avants, 2012) and other packages such as MindBoggle,2 Nipype3 (Gorgolewski et al., 2011), and the Advanced Normalization Tools (ANTs),4 (Avants et al., 2010, 2011) we notice an increasing number of publications that intend a fair comparison of algorithms which, in principle, is a good thing. Our concern is the lack of detail with which these comparisons are often presented and the corresponding possibility of instrumentation bias (Sackett, 1979) where “defects in the calibration or maintenance of measurement instruments may lead to systematic deviations from true values” (considering software as a type of instrument requiring proper “calibration” and “maintenance” for accurate measurements). Based on our experience (including our own mistakes), we propose a preliminary set of guidelines that seek to minimize such bias with the understanding that the discussion will require a more comprehensive response from the larger neuroscience community. Our intent is to raise awareness in both authors and reviewers to issues that arise when comparing quantitative algorithms. Although herein we focus largely on image registration, these recommendations are relevant for other application areas in biologically-focused computational image analysis, and for reproducible computational science in general. This commentary complements recent papers that highlight statistical bias (Kriegeskorte et al., 2009; Vul and Pashler, 2012), bias induced by registration metrics (Tustison et al., 2012), and registration strategy (Yushkevich et al., 2010) and guideline papers for software development (Prlic and Procter, 2012).
The degree to which one identifies as male or female has a profound impact on one's life. Yet, there is a limited understanding of what contributes to this important characteristic termed gender identity. In order to reveal factors influencing gender identity, studies have focused on people who report strong feelings of being the opposite sex, such as male-to-female (MTF) transsexuals.
To investigate potential neuroanatomical variations associated with transsexualism, we compared the regional thickness of the cerebral cortex between 24 MTF transsexuals who had not yet been treated with cross-sex hormones and 24 age-matched control males.
Results revealed thicker cortices in MTF transsexuals, both within regions of the left hemisphere (i.e., frontal and orbito-frontal cortex, central sulcus, perisylvian regions, paracentral gyrus) and right hemisphere (i.e., pre-/post-central gyrus, parietal cortex, temporal cortex, precuneus, fusiform, lingual, and orbito-frontal gyrus).
These findings provide further evidence that brain anatomy is associated with gender identity, where measures in MTF transsexuals appear to be shifted away from gender-congruent men.
We introduce the Mindboggle-101 dataset, the largest and most complete set of free, publicly accessible, manually labeled human brain images. To manually label the macroscopic anatomy in magnetic resonance images of 101 healthy participants, we created a new cortical labeling protocol that relies on robust anatomical landmarks and minimal manual edits after initialization with automated labels. The “Desikan–Killiany–Tourville” (DKT) protocol is intended to improve the ease, consistency, and accuracy of labeling human cortical areas. Given how difficult it is to label brains, the Mindboggle-101 dataset is intended to serve as brain atlases for use in labeling other brains, as a normative dataset to establish morphometric variation in a healthy population for comparison against clinical populations, and contribute to the development, training, testing, and evaluation of automated registration and labeling algorithms. To this end, we also introduce benchmarks for the evaluation of such algorithms by comparing our manual labels with labels automatically generated by probabilistic and multi-atlas registration-based approaches. All data and related software and updated information are available on the http://mindboggle.info/data website.
Multi-atlas segmentation is an effective approach for automatically labeling objects of interest in biomedical images. In this approach, multiple expert-segmented example images, called \emph{atlases}, are registered to a target image, and deformed atlas segmentations are combined using \emph{label fusion}. Among the proposed label fusion strategies, weighted voting with spatially varying weight distributions derived from atlas-target intensity similarity have been particularly successful. However, one limitation of these strategies is that the weights are computed independently for each atlas, without taking into account the fact that different atlases may produce similar label errors. To address this limitation, we propose a new solution for the label fusion problem, in which weighted voting is formulated in terms of minimizing the total expectation of labeling error, and in which pairwise dependency between atlases is explicitly modeled as the joint probability of two atlases making a segmentation error at a voxel. This probability is approximated using intensity similarity between a pair of atlases and the target image in the neighborhood of each voxel. We validate our method in two medical image segmentation problems: hippocampus segmentation and hippocampus subfield segmentation in magnetic resonance (MR) images. For both problems, we show consistent and significant improvement over label fusion strategies that assign atlas weights independently.
FreeSurfer is a popular software package to measure cortical thickness and volume of neuroanatomical structures. However, little if any is known about measurement reliability across various data processing conditions. Using a set of 30 anatomical T1-weighted 3T MRI scans, we investigated the effects of data processing variables such as FreeSurfer version (v4.3.1, v4.5.0, and v5.0.0), workstation (Macintosh and Hewlett-Packard), and Macintosh operating system version (OSX 10.5 and OSX 10.6). Significant differences were revealed between FreeSurfer version v5.0.0 and the two earlier versions. These differences were on average 8.8±6.6% (range 1.3–64.0%) (volume) and 2.8±1.3% (1.1–7.7%) (cortical thickness). About a factor two smaller differences were detected between Macintosh and Hewlett-Packard workstations and between OSX 10.5 and OSX 10.6. The observed differences are similar in magnitude as effect sizes reported in accuracy evaluations and neurodegenerative studies.
The main conclusion is that in the context of an ongoing study, users are discouraged to update to a new major release of either FreeSurfer or operating system or to switch to a different type of workstation without repeating the analysis; results thus give a quantitative support to successive recommendations stated by FreeSurfer developers over the years. Moreover, in view of the large and significant cross-version differences, it is concluded that formal assessment of the accuracy of FreeSurfer is desirable.
Voxel-based morphometry (VBM) studies have interpreted longitudinal medication- or behaviorally induced changes observed on T1-weighted magnetic resonance images (MRIs) as changes in neuronal structure. Although neurogenesis or atrophy certainly occurs, the use of T1-weighted scans to identify change in brain structure in vivo in humans has vulnerability: the T1 relaxation time for arterial blood and gray matter are not clearly distinguishable and therefore, apparent reported structural findings might be at least partially related to changes in blood flow or other physiological signals. To examine the hypothesis that apparent structural modifications may reflect changes introduced by additional mechanisms irrespective of potential neuronal growth/atrophy, we acquired a high-resolution T1-weighted structural scan and a 5-min perfusion fMRI scan (a measurement of blood flow), before and after administration of an acute pharmacological manipulation. In a within-subject design, 15 subjects were either un-medicated or were administered a 20 mg dose of baclofen (an FDA-approved anti-spastic) approximately 110 min before acquiring a T1-weighted scan and a pseudo continuous arterial spin labeled perfusion fMRI scan. Using diffeomorphic anatomical registration through exponentiated lie algebra within SPM7, we observed macroscopic, and therefore implausible, baclofen-induced decreases in VBM 'gray matter' signal in the dorsal rostral anterior cingulate (family wise error corrected at p < 0.04, T = 6.54, extent: 1,460 voxels) that overlapped with changes in blood flow. Given that gray matter reductions are unlikely following a single dose of medication these findings suggest that changes in blood flow are masquerading as reductions in gray matter on the T1-weighted scan irrespective of the temporal interval between baseline measures and longitudinal manipulations. These results underscore the crucial and immediate need to develop in vivo neuroimaging biomarkers for humans that can uniquely capture changes in neuronal structure dissociable from those related to blood flow or other physiological signals.
Alzheimer's disease (AD) is associated with neurodegeneration in vulnerable limbic and heteromodal regions of the cerebral
cortex, detectable in vivo using magnetic resonance imaging. It is not clear whether abnormalities of cortical anatomy in
AD can be reliably measured across different subject samples, how closely they track symptoms, and whether they are detectable
prior to symptoms. An exploratory map of cortical thinning in mild AD was used to define regions of interest that were applied
in a hypothesis-driven fashion to other subject samples. Results demonstrate a reliably quantifiable in vivo signature of
abnormal cortical anatomy in AD, which parallels known regional vulnerability to AD neuropathology. Thinning in vulnerable
cortical regions relates to symptom severity even in the earliest stages of clinical symptoms. Furthermore, subtle thinning
is present in asymptomatic older controls with brain amyloid binding as detected with amyloid imaging. The reliability and
clinical validity of AD-related cortical thinning suggests potential utility as an imaging biomarker. This “disease signature”
approach to cortical morphometry, in which disease effects are mapped across the cortical mantle and then used to define ROIs
for hypothesis-driven analyses, may provide a powerful methodological framework for studies of neuropsychiatric diseases.
We introduce Atropos, an ITK-based multivariate n-class open source segmentation algorithm distributed with ANTs ( http://www.picsl.upenn.edu/ANTs). The Bayesian formulation of the segmentation problem is solved using the Expectation Maximization (EM) algorithm with the modeling of the class intensities based on either parametric or non-parametric finite mixtures. Atropos is capable of incorporating spatial prior probability maps (sparse), prior label maps and/or Markov Random Field (MRF) modeling. Atropos has also been efficiently implemented to handle large quantities of possible labelings (in the experimental section, we use up to 69 classes) with a minimal memory footprint. This work describes the technical and implementation aspects of Atropos and evaluates its performance on two different ground-truth datasets. First, we use the BrainWeb dataset from Montreal Neurological Institute to evaluate three-tissue segmentation performance via (1) K-means segmentation without use of template data; (2) MRF segmentation with initialization by prior probability maps derived from a group template; (3) Prior-based segmentation with use of spatial prior probability maps derived from a group template. We also evaluate Atropos performance by using spatial priors to drive a 69-class EM segmentation problem derived from the Hammers atlas from University College London. These evaluation studies, combined with illustrative examples that exercise Atropos options, demonstrate both performance and wide applicability of this new platform-independent open source segmentation tool.
Sports experts represent a population of people who have acquired expertise in sports training and competition. Recently, the number of studies on sports experts has increased; however, neuroanatomical changes following extensive training are not fully understood. In this study, we used cortical thickness measurement to investigate the brain anatomical characteristics of professional divers with extensive training experience. A comparison of the brain anatomical characteristics of the non-athlete group with those of the athlete group revealed three regions with significantly increased cortical thickness in the athlete group. These regions included the left superior temporal sulcus, the right orbitofrontal cortex and the right parahippocampal gyrus. Moreover, a significant positive correlation between the mean cortical thickness of the right parahippocampal gyrus and the training experience was detected, which might indicate the effect of extensive training on diving players' brain structure.
Measuring the entorhinal cortex (ERC) is challenging due to lateral border discrimination from the perirhinal cortex. From a sample of 39 nondemented older adults who completed volumetric image scans and verbal memory indices, we examined reliability and validity concerns for three ERC protocols with different lateral boundary guidelines (i.e., Goncharova, Dickerson, Stoub, & deToledo-Morrell, 2001; Honeycutt et al., 1998; Insausti et al., 1998). We used three novice raters to assess inter-rater reliability on a subset of scans (216 total ERCs), with the entire dataset measured by one rater with strong intra-rater reliability on each technique (234 total ERCs). We found moderate to strong inter-rater reliability for two techniques with consistent ERC lateral boundary endpoints (Goncharova, Honeycutt), with negligible to moderate reliability for the technique requiring consideration of collateral sulcal depth (Insausti). Left ERC and story memory associations were moderate and positive for two techniques designed to exclude the perirhinal cortex (Insausti, Goncharova), with the Insausti technique continuing to explain 10% of memory score variance after additionally controlling for depression symptom severity. Right ERC-story memory associations were nonexistent after excluding an outlier. Researchers are encouraged to consider challenges of rater training for ERC techniques and how lateral boundary endpoints may impact structure-function associations. (JINS, 2010, 16, 846-855.).
Neuro-imaging studies demonstrate plasticity of cortical gray matter before and after practice for some motor and cognitive tasks in adults. Other imaging studies show functional changes after practice, but there is not yet direct evidence of how structural and functional changes may be related. A fundamental question is whether they occur at the same cortical sites, adjacent sites, or sites in other parts of a network.
Using a 3 T MRI, we obtained structural and functional images in adolescent girls before and after practice on a visual-spatial problem-solving computer game, Tetris. After three months of practice, compared to the structural scans of controls, the group with Tetris practice showed thicker cortex, primarily in two areas: left BAs 6 and 22/38. Based on fMRI BOLD signals, the Tetris group showed cortical activations throughout the brain while playing Tetris, but significant BOLD decreases, mostly in frontal areas, were observed after practice. None of these BOLD decreases, however, overlapped with the cortical thickness changes.
Regional cortical thickness changes were observed after three months of Tetris practice. Over the same period, brain activity decreases were observed in several other areas. These data indicate that structural change in one brain area does not necessarily result in functional change in the same location, at least on the levels assessed with these MRI methods.
Reliability coefficients often take the form of intraclass correlation coefficients. In this article, guidelines are given for choosing among 6 different forms of the intraclass correlation for reliability studies in which n targets are rated by k judges. Relevant to the choice of the coefficient are the appropriate statistical model for the reliability study and the applications to be made of the reliability results. Confidence intervals for each of the forms are reviewed. (23 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved).
A novel approach to correcting for intensity nonuniformity in magnetic resonance (MR) data is described that achieves high performance without requiring a model of the tissue classes present. The method has the advantage that it can be applied at an early stage in an automated data analysis, before a tissue model is available. Described as nonparametric nonuniform intensity normalization (N3), the method is independent of pulse sequence and insensitive to pathological data that might otherwise violate model assumptions. To eliminate the dependence of the field estimate on anatomy, an iterative approach is employed to estimate both the multiplicative bias field and the distribution of the true tissue intensities. The performance of this method is evaluated using both real and simulated MR data.
The cortex is the outermost thin layer of gray matter in the brain; geometric measurement of the cortex helps in understanding brain anatomy and function. In the quantitative analysis of the cortex from MR images, extracting the structure and obtaining a representation for various measurements are key steps. While manual segmentation is tedious and labor intensive, automatic reliable efficient segmentation and measurement of the cortex remain challenging problems, due to its convoluted nature. Here we present a new approach of coupled-surfaces propagation, using level set methods to address such problems. Our method is motivated by the nearly constant thickness of the cortical mantle and takes this tight coupling as an important constraint. By evolving two embedded surfaces simultaneously, each driven by its own image-derived information while maintaining the coupling, a final representation of the cortical bounding surfaces and an automatic segmentation of the cortex are achieved. Characteristics of the cortex, such as cortical surface area, surface curvature, and cortical thickness, are then evaluated. The level set implementation of surface propagation offers the advantage of easy initialization, computational efficiency, and the ability to capture deep sulcal folds. Results and validation from various experiments on both simulated and real three-dimensional (3-D) MR images are provided.
Accurate and automated methods for measuring the thickness of human cerebral cortex could provide powerful tools for diagnosing and studying a variety of neurodegenerative and psychiatric disorders. Manual methods for estimating cortical thickness from neuroimaging data are labor intensive, requiring several days of effort by a trained anatomist. Furthermore, the highly folded nature of the cortex is problematic for manual techniques, frequently resulting in measurement errors in regions in which the cortical surface is not perpendicular to any of the cardinal axes. As a consequence, it has been impractical to obtain accurate thickness estimates for the entire cortex in individual subjects, or group statistics for patient or control populations. Here, we present an automated method for accurately measuring the thickness of the cerebral cortex across the entire brain and for generating cross-subject statistics in a coordinate system based on cortical anatomy. The intersubject standard deviation of the thickness measures is shown to be less than 0.5 mm, implying the ability to detect focal atrophy in small populations or even individual subjects. The reliability and accuracy of this new method are assessed by within-subject test-retest studies, as well as by comparison of cross-subject regional thickness measures with published values.
A novel approach to correcting for intensity nonuniformity in magnetic resonance (MR) data is described that achieves high performance without requiring a model of the tissue classes present. The method has the advantage that it can be applied at an early stage in an automated data analysis, before a tissue model is available. Described as nonparametric nonuniform intensity normalization (N3), the method is independent of pulse sequence and insensitive to pathological data that might otherwise violate model assumptions. To eliminate the dependence of the field estimate on anatomy, an iterative approach is employed to estimate both the multiplicative bias field and the distribution of the true tissue intensities. The performance of this method is evaluated using both real and simulated MR data.
We present a technique for automatically assigning a neuroanatomical label to each location on a cortical surface model based on probabilistic information estimated from a manually labeled training set. This procedure incorporates both geometric information derived from the cortical model, and neuroanatomical convention, as found in the training set. The result is a complete labeling of cortical sulci and gyri. Examples are given from two different training sets generated using different neuroanatomical conventions, illustrating the flexibility of the algorithm. The technique is shown to be comparable in accuracy to manual labeling.
Random forests are a combination of tree predictors such that each tree depends on the values of a random vector sampled independently and with the same distribution for all trees in the forest. The generalization error for forests converges a.s. to a limit as the number of trees in the forest becomes large. The generalization error of a forest of tree classifiers depends on the strength of the individual trees in the forest and the correlation between them. Using a random selection of features to split each node yields error rates that compare favorably to Adaboost (Y. Freund & R. Schapire, Machine Learning: Proceedings of the Thirteenth International conference, ***, 148–156), but are more robust with respect to noise. Internal estimates monitor error, strength, and correlation and these are used to show the response to increasing the number of features used in the splitting. Internal estimates are also used to measure variable importance. These ideas are also applicable to regression.
Large-scale longitudinal multi-site MRI brain morphometry studies are becoming increasingly crucial to characterize both normal and clinical population groups using fully automated segmentation tools. The test–retest reproducibility of morphometry data acquired across multiple scanning sessions, and for different MR vendors, is an important reliability indicator since it defines the sensitivity of a protocol to detect longitudinal effects in a consortium. There is very limited knowledge about how across-session reliability of morphometry estimates might be affected by different 3 T MRI systems. Moreover, there is a need for optimal acquisition and analysis protocols in order to reduce sample sizes. A recent study has shown that the longitudinal FreeSurfer segmentation offers improved within session test–retest reproducibility relative to the cross-sectional segmentation at one 3 T site using a nonstandard multi-echo MPRAGE sequence. In this study we implement a multi-site 3 T MRI morphometry protocol based on vendor provided T1 structural sequences from different vendors (3D MPRAGE on Siemens and Philips, 3D IR-SPGR on GE) implemented in 8 sites located in 4 European countries. The protocols used mild acceleration factors (1.5–2) when possible. We acquired across-session test–retest structural data of a group of healthy elderly subjects (5 subjects per site) and compared the across-session reproducibility of two full-brain automated segmentation methods based on either longitudinal or cross-sectional FreeSurfer processing. The segmentations include cortical thickness, intracranial, ventricle and subcortical volumes. Reproducibility is evaluated as absolute changes relative to the mean (%), Dice coefficient for volume overlap and intraclass correlation coefficients across two sessions. We found that this acquisition and analysis protocol gives comparable reproducibility results to previous studies that used longer acquisitions without acceleration. We also show that the longitudinal processingis systematically more reliable across sites regardless of MRI system differences. The reproducibility errors of the longitudinal segmentations are on average approximately half of those obtained with the cross sectional analysis for all volume segmentations and for entorhinal cortical thickness. No significant differences in reliability are found between the segmentation methods for the other cortical thickness estimates. The average of two MPRAGE volumes acquired within each test–retest session did not systematically improve the across-session reproducibility of morphometry estimates. Our results extend those from previous studies that showed improved reliability of the longitudinal analysis at single sites and/or with non-standard acquisition methods. The multi-site acquisition and analysis protocol presented here is promising for clinical applications since it allows for smaller sample sizes per MRI site or shorter trials in studies evaluating the role of potential biomarkers to predict disease progression or treatment effects.
The analysis of neuroimaging data from pediatric populations presents several challenges. There are normal variations in brain shape from infancy to adulthood and normal developmental changes related to tissue maturation. Measurement of cortical thickness is one important way to analyze such developmental tissue changes. We developed a novel framework that allows group-wise automatic mesh-based analysis of cortical thickness. Our approach is divided into four main parts. First an individual pre-processing pipeline is applied on each subject to create genus-zero inflated white matter cortical surfaces with cortical thickness measurements. The second part performs an entropy-based group-wise shape correspondence on these meshes using a particle system, which establishes a trade-off between an even sampling of the cortical surfaces and the similarity of corresponding points across the population using sulcal depth information and spatial proximity. A novel automatic initial particle sampling is performed using a matched 98-lobe parcellation map prior to a particle-splitting phase. Third, corresponding re-sampled surfaces are computed with interpolated cortical thickness measurements, which are finally analyzed via a statistical vertex-wise analysis module. This framework consists of a pipeline of automated 3D Slicer compatible modules. It has been tested on a small pediatric dataset and incorporated in an open-source C++ based high-level module called GAMBIT. GAMBIT's setup allows efficient batch processing, grid computing and quality control. The current research focuses on the use of an average template for correspondence and surface re-sampling, as well as thorough validation of the framework and its application to clinical pediatric studies.
We present a technique for automatically assigning a neuroanatomical label to each voxel in an MRI volume based on probabilistic information automatically estimated from a manually labeled training set. In contrast to existing segmentation procedures that only label a small number of tissue classes, the current method assigns one of 37 labels to each voxel, including left and right caudate, putamen, pallidum, thalamus, lateral ventricles, hippocampus, and amygdala. The classification technique employs a registration procedure that is robust to anatomical variability, including the ventricular enlargement typically associated with neurological diseases and aging. The technique is shown to be comparable in accuracy to manual labeling, and of sufficient sensitivity to robustly detect changes in the volume of noncortical structures that presage the onset of probable Alzheimer's disease.
Purpose:
The analysis of the human cerebral cortex and the measurement of its thickness based on MRI data can provide insight into normal brain development and neurodegenerative disorders. Accurate and reproducible results of the cortical thickness measurement are desired for sensitive detection. This study compares ultra-high resolution data acquired at 7T with 3T data for determination of the cortical thickness of the human brain. The impact of field strength, resolution, and processing method is evaluated systematically.
Methods:
Five subjects were scanned at 3T (1 mm isotropic resolution) and 7T (1 mm and 0.5 mm isotropic resolution) with 3D MP-RAGE and 3D gradient echo methods. The inhomogeneous signal and contrast of the 7T data due to the B1 field was corrected by division of the MP-RAGE with the GE. ARCTIC, utilizing a voxel-based approach, and FreeSurfer, utilizing a surface-based approach, have been used to compute the cortical thickness of the high resolution 3T and 7T data and of the ultra-high resolution 7T data. FreeSurfer is not designed to process data with a spatial resolution other than 1mm and was modified to avoid this limitation. Additionally SPM and FSL have been used to generate segmentations which were further processed with ARCTIC to determine the cortical thickness.
Results and conclusion:
At identical resolution, the cortical thickness determination yielded consistent results between 3T and 7T confirming the robustness of the acquisition and processing against potential field strength related effects. However, the ultra-high resolution 7T data resulted in significantly reduced values for the cortical thickness estimation compared to the lower resolution data. The reduction in thickness amounts approximately one sixth to one third, depending on the processing algorithm and software used. This suggests a bias in the gray matter segmentation due to partial volume effects and indicates that true cortical thickness is overestimated by most current MR studies using both a voxel-based or surface-based method and can be more accurately determined with high resolution imaging at 7T.
Older adults exhibit global reductions in cortical surface area, but little is known about the regional patterns of reductions or how these relate to other measures of brain structure. This knowledge is critical to understanding the dynamic relationship between different macrostructural properties of the cortex throughout adult life. Here, cortical arealization, local gyrification index (LGI), and cortical thickness were measured vertex wise across the brain surface in 322 healthy adults (20-85 years), with the aims of 1) characterizing age patterns of the three separate cortical measures and 2) testing the age-independent relationships among cortical surface area, gyrification, and thickness. Surface area showed strong age-related decreases, particularly pronounced in dorsomedial prefrontal, lateral temporal, and fusiform cortices, independently of total white matter volume. LGI decreased with age independently of regional surface area, with strongest effects laterally, extending from the angular gyrus in all directions. As expected, regional surface area and LGI were positively related. However, both measures correlated negatively with thickness, indicating increasing local arealization and gyrification with decreasing cortical thickness. We suggest that this pattern of regional "cortical stretching" reflects the well-established phylogenetic principle of maximizing surface area and gyrification rather than increase thickness to facilitate brain connectivity and functional development.
We present a novel, computerized method of examining cerebral cortical thickness. The normal cortex varies in thickness from 2 to 4 mm, reflecting the morphology of neuronal sublayers. Cortical pathologies often manifest abnormal variations in thickness, with examples of Alzheimer's disease and cortical dysplasia as thin and thick cortex, respectively. Radiologically, images are 2-D slices through a highly convoluted 3-D object. Depending on the relative orientation of the slices with respect to the object, it is impossible to deduce abnormal cortical thickness without additional information from neighboring slices. We approach the problem by applying Laplace's Equation (∇2ψ = 0) from mathematical physics. The volume of the cortex is represented as the domain for the solution of the differential equation, with separate boundary conditions at the gray-white junction and the gray-CSF junction. Normalized gradients of ψ form a vector field, representing tangent vectors along field lines connecting both boundaries. We define the cortical thickness at any point in the cortex to be the pathlength along such lines. Key advantages of this method are that it is fully three-dimensional, and the thickness is uniquely defined for any point in the cortex. We present graphical results that map cortical thickness everywhere in a normal brain. Results show global variations in cortical thickness consistent with known neuroanatomy. The application of this technique to visualization of cortical thickness in brains with known pathology has broad clinical implications. Hum. Brain Mapping 11:12–32, 2000. © 2000 Wiley-Liss, Inc.
A variant of the popular nonparametric nonuniform intensity normalization (N3) algorithm is proposed for bias field correction. Given the superb performance of N3 and its public availability, it has been the subject of several evaluation studies. These studies have demonstrated the importance of certain parameters associated with the B -spline least-squares fitting. We propose the substitution of a recently developed fast and robust B-spline approximation routine and a modified hierarchical optimization scheme for improved bias field correction over the original N3 algorithm. Similar to the N3 algorithm, we also make the source code, testing, and technical documentation of our contribution, which we denote as ??N4ITK,?? available to the public through the Insight Toolkit of the National Institutes of Health. Performance assessment is demonstrated using simulated data from the publicly available Brainweb database, hyperpolarized 3He lung image data, and 9.4T postmortem hippocampus data.
Automatic computer processing of large multidimensional images such as those produced by magnetic resonance imaging (MRI) is greatly aided by deformable models, which are used to extract, identify, and quantify specific neuroanatomic structures. A general method of deforming polyhedra is presented here, with two novel features. First, explicit prevention of self-intersecting surface geometries is provided, unlike conventional deformable models, which use regularization constraints to discourage but not necessarily prevent such behavior. Second, deformation of multiple surfaces with intersurface proximity constraints allows each surface to help guide other surfaces into place using model-based constraints such as expected thickness of an anatomic surface. These two features are used advantageously to identify automatically the total surface of the outer and inner boundaries of cerebral cortical gray matter from normal human MR images, accurately locating the depths of the sulci, even where noise and partial volume artifacts in the image obscure the visibility of sulci. The extracted surfaces are enforced to be simple two-dimensional manifolds (having the topology of a sphere), even though the data may have topological holes. This automatic 3-D cortex segmentation technique has been applied to 150 normal subjects, simultaneously extracting both the gray/white and gray/cerebrospinal fluid interface from each individual. The collection of surfaces has been used to create a spatial map of the mean and standard deviation for the location and the thickness of cortical gray matter. Three alternative criteria for defining cortical thickness at each cortical location were developed and compared. These results are shown to corroborate published postmortem and in vivo measurements of cortical thickness.
Longitudinal image analysis has become increasingly important in clinical studies of normal aging and neurodegenerative disorders. Furthermore, there is a growing appreciation of the potential utility of longitudinally acquired structural images and reliable image processing to evaluate disease modifying therapies. Challenges have been related to the variability that is inherent in the available cross-sectional processing tools, to the introduction of bias in longitudinal processing and to potential over-regularization. In this paper we introduce a novel longitudinal image processing framework, based on unbiased, robust, within-subject template creation, for automatic surface reconstruction and segmentation of brain MRI of arbitrarily many time points. We demonstrate that it is essential to treat all input images exactly the same as removing only interpolation asymmetries is not sufficient to remove processing bias. We successfully reduce variability and avoid over-regularization by initializing the processing in each time point with common information from the subject template. The presented results show a significant increase in precision and discrimination power while preserving the ability to detect large anatomical deviations; as such they hold great potential in clinical applications, e.g. allowing for smaller sample sizes or shorter trials to establish disease specific biomarkers or to quantify drug effects.
The paper describes the symbolic notation and syntax for specifying factorial models for analysis of variance in the control language of the GENSTAT statistical program system at Rothamsted. The notation generalizes that of Nelder (1965). Algorith AS 65 (Rogers. 1973) converts factorial model formulas in this notation to a list of model terms represented as binary integers. A further extension of the syntax is discussed for specifying models generally (including non-linear forms).
A nonparametric and unsupervised method of automatic threshold selection for picture segmentation is presented. An otpimal threshold is selected by the discriminant criterion, namely, so as the maximize the separability of the resultant classes in gray levels. The procedure is very simple, utilizing only the zeroth- and first-order cumulative moments of the gray-level histogram. It is strightforward to extend the method to multithreshold problems. Several experimental results are also presented to support the validity of the method.
Cortical thickness estimation performed in-vivo via magnetic resonance imaging is an important technique for the diagnosis and understanding of the progression of neurodegenerative diseases. Currently, two different computational paradigms exist, with methods generally classified as either surface or voxel-based. This paper provides a much needed comparison of the surface-based method FreeSurfer and two voxel-based methods using clinical data. We test the effects of computing regional statistics using two different atlases and demonstrate that this makes a significant difference to the cortical thickness results. We assess reproducibility, and show that FreeSurfer has a regional standard deviation of thickness difference on same day scans that is significantly lower than either a Laplacian or Registration based method and discuss the trade off between reproducibility and segmentation accuracy caused by bending energy constraints. We demonstrate that voxel-based methods can detect similar patterns of group-wise differences as well as FreeSurfer in typical applications such as producing group-wise maps of statistically significant thickness change, but that regional statistics can vary between methods. We use a Support Vector Machine to classify patients against controls and did not find statistically significantly different results with voxel based methods compared to FreeSurfer. Finally we assessed longitudinal performance and concluded that currently FreeSurfer provides the most plausible measure of change over time, with further work required for voxel based methods.
Idiopathic Parkinson's disease (PD) is a neurodegenerative disorder diagnosed on the basis of motor symptoms, but that also includes cognitive and visuo-spatial deficits. Though PD is known to initially affect subcortical regions, the cortex also exhibits neuronal loss in the course of the disease as post mortem studies have shown. So far, PD-related pattern of cortical damage remains unclear, because of disease-caused heterogeneity, and also in part because of methodological issues such as the limitations of Voxel Based Morphometry. Here corticometry was used, a technique that decouples local surface from thickness, to obtain a better picture of PD corticomorphometric patterns.
Modern MRI image processing methods have yielded quantitative, morphometric, functional, and structural assessments of the human brain. These analyses typically exploit carefully optimized protocols for specific imaging targets. Algorithm investigators have several excellent public data resources to use to test, develop, and optimize their methods. Recently, there has been an increasing focus on combining MRI protocols in multi-parametric studies. Notably, these have included innovative approaches for fusing connectivity inferences with functional and/or anatomical characterizations. Yet, validation of the reproducibility of these interesting and novel methods has been severely hampered by the limited availability of appropriate multi-parametric data. We present an imaging protocol optimized to include state-of-the-art assessment of brain function, structure, micro-architecture, and quantitative parameters within a clinically feasible 60-min protocol on a 3-T MRI scanner. We present scan-rescan reproducibility of these imaging contrasts based on 21 healthy volunteers (11 M/10 F, 22-61 years old). The cortical gray matter, cortical white matter, ventricular cerebrospinal fluid, thalamus, putamen, caudate, cerebellar gray matter, cerebellar white matter, and brainstem were identified with mean volume-wise reproducibility of 3.5%. We tabulate the mean intensity, variability, and reproducibility of each contrast in a region of interest approach, which is essential for prospective study planning and retrospective power analysis considerations. Anatomy was highly consistent on structural acquisition (~1-5% variability), while variation on diffusion and several other quantitative scans was higher (~<10%). Some sequences are particularly variable in specific structures (ASL exhibited variation of 28% in the cerebral white matter) or in thin structures (quantitative T2 varied by up to 73% in the caudate) due, in large part, to variability in automated ROI placement. The richness of the joint distribution of intensities across imaging methods can be best assessed within the context of a particular analysis approach as opposed to a summary table. As such, all imaging data and analysis routines have been made publicly and freely available. This effort provides the neuroimaging community with a resource for optimization of algorithms that exploit the diversity of modern MRI modalities. Additionally, it establishes a baseline for continuing development and optimization of multi-parametric imaging protocols.
The United States National Institutes of Health (NIH) commit significant support to open-source data and software resources in order to foment reproducibility in the biomedical imaging sciences. Here, we report and evaluate a recent product of this commitment: Advanced Neuroimaging Tools (ANTs), which is approaching its 2.0 release. The ANTs open source software library consists of a suite of state-of-the-art image registration, segmentation and template building tools for quantitative morphometric analysis. In this work, we use ANTs to quantify, for the first time, the impact of similarity metrics on the affine and deformable components of a template-based normalization study. We detail the ANTs implementation of three similarity metrics: squared intensity difference, a new and faster cross-correlation, and voxel-wise mutual information. We then use two-fold cross-validation to compare their performance on openly available, manually labeled, T1-weighted MRI brain image data of 40 subjects (UCLA's LPBA40 dataset). We report evaluation results on cortical and whole brain labels for both the affine and deformable components of the registration. Results indicate that the best ANTs methods are competitive with existing brain extraction results (Jaccard=0.958) and cortical labeling approaches. Mutual information affine mapping combined with cross-correlation diffeomorphic mapping gave the best cortical labeling results (Jaccard=0.669±0.022). Furthermore, our two-fold cross-validation allows us to quantify the similarity of templates derived from different subgroups. Our open code, data and evaluation scripts set performance benchmark parameters for this state-of-the-art toolkit. This is the first study to use a consistent transformation framework to provide a reproducible evaluation of the isolated effect of the similarity metric on optimal template construction and brain labeling.
Establishing correspondences across brains for the purposes of comparison and group analysis is almost universally done by registering images to one another either directly or via a template. However, there are many registration algorithms to choose from. A recent evaluation of fully automated nonlinear deformation methods applied to brain image registration was restricted to volume-based methods. The present study is the first that directly compares some of the most accurate of these volume registration methods with surface registration methods, as well as the first study to compare registrations of whole-head and brain-only (de-skulled) images. We used permutation tests to compare the overlap or Hausdorff distance performance for more than 16,000 registrations between 80 manually labeled brain images. We compared every combination of volume-based and surface-based labels, registration, and evaluation. Our primary findings are the following: 1. de-skulling aids volume registration methods; 2. custom-made optimal average templates improve registration over direct pairwise registration; and 3. resampling volume labels on surfaces or converting surface labels to volumes introduces distortions that preclude a fair comparison between the highest ranking volume and surface registration methods using present resampling methods. From the results of this study, we recommend constructing a custom template from a limited sample drawn from the same or a similar representative population, using the same algorithm used for registering brains to the template.
We evaluate the impact of template choice on template-based segmentation of the hippocampus in epilepsy. Four dataset-specific strategies are quantitatively contrasted: the "closest to average" individual template, the average shape version of the closest to average template, a best appearance template and the best appearance and shape template proposed here and implemented in the open source toolkit Advanced Normalization Tools (ANTS). The cross-correlation similarity metric drives the correspondence model and is used consistently to determine the optimal appearance. Minimum shape distance in the diffeomorphic space determines optimal shape. Our evaluation results show that, with respect to gold-standard manual labeling of hippocampi in epilepsy, optimal shape and appearance template construction outperforms the other strategies for gaining data-derived templates. Our results also show the improvement is most significant on the diseased side and insignificant on the healthy side. Thus, the importance of the template increases when used to study pathology and may be less critical for normal control studies. Furthermore, explicit geometric optimization of the shape component of the unbiased template positively impacts the study of diseased hippocampi.
Obesity is associated with increased risk for cardiovascular health problems including diabetes, hypertension, and stroke. These cardiovascular afflictions increase risk for cognitive decline and dementia, but it is unknown whether these factors, specifically obesity and Type II diabetes, are associated with specific patterns of brain atrophy. We used tensor-based morphometry (TBM) to examine gray matter (GM) and white matter (WM) volume differences in 94 elderly subjects who remained cognitively normal for at least 5 years after their scan. Bivariate analyses with corrections for multiple comparisons strongly linked body mass index (BMI), fasting plasma insulin (FPI) levels, and Type II Diabetes Mellitus (DM2) with atrophy in frontal, temporal, and subcortical brain regions. A multiple regression model, also correcting for multiple comparisons, revealed that BMI was still negatively correlated with brain atrophy (FDR <5%), while DM2 and FPI were no longer associated with any volume differences. In an Analysis of Covariance (ANCOVA) model controlling for age, gender, and race, obese subjects with a high BMI (BMI > 30) showed atrophy in the frontal lobes, anterior cingulate gyrus, hippocampus, and thalamus compared with individuals with a normal BMI (18.5-25). Overweight subjects (BMI: 25-30) had atrophy in the basal ganglia and corona radiata of the WM. Overall brain volume did not differ between overweight and obese persons. Higher BMI was associated with lower brain volumes in overweight and obese elderly subjects. Obesity is therefore associated with detectable brain volume deficits in cognitively normal elderly subjects.
Cortical thickness is an important biomarker for image-based studies of the brain. A diffeomorphic registration based cortical thickness (DiReCT) measure is introduced where a continuous one-to-one correspondence between the gray matter-white matter interface and the estimated gray matter-cerebrospinal fluid interface is given by a diffeomorphic mapping in the image space. Thickness is then defined in terms of a distance measure between the interfaces of this sheet like structure. This technique also provides a natural way to compute continuous estimates of thickness within buried sulci by preventing opposing gray matter banks from intersecting. In addition, the proposed method incorporates neuroanatomical constraints on thickness values as part of the mapping process. Evaluation of this method is presented on synthetic images. As an application to brain images, a longitudinal study of thickness change in frontotemporal dementia (FTD) spectrum disorder is reported.
Several algorithms for measuring the cortical thickness in the human brain from MR image volumes have been described in the literature, the majority of which rely on fitting deformable models to the inner and outer cortical surfaces. However, the constraints applied during the model fitting process in order to enforce spherical topology and to fit the outer cortical surface in narrow sulci, where the cerebrospinal fluid (CSF) channel may be obscured by partial voluming, may introduce bias in some circumstances, and greatly increase the processor time required. In this paper we describe an alternative, voxel based technique that measures the cortical thickness using inversion recovery anatomical MR images. Grey matter, white matter and CSF are identified through segmentation, and edge detection is used to identify the boundaries between these tissues. The cortical thickness is then measured along the local 3D surface normal at every voxel on the inner cortical surface. The method was applied to 119 normal volunteers, and validated through extensive comparisons with published measurements of both cortical thickness and rate of thickness change with age. We conclude that the proposed technique is generally faster than deformable model-based alternatives, and free from the possibility of model bias, but suffers no reduction in accuracy. In particular, it will be applicable in data sets showing severe cortical atrophy, where thinning of the gyri leads to points of high curvature, and so the fitting of deformable models is problematic.
In both diagnostic and research applications, the interpretation of MR images of the human brain is facilitated when different data sets can be compared by visual inspection of equivalent anatomical planes. Quantitative analysis with predefined atlas templates often requires the initial alignment of atlas and image planes. Unfortunately, the axial planes acquired during separate scanning sessions are often different in their relative position and orientation, and these slices are not coplanar with those in the atlas. We have developed a completely automatic method to register a given volumetric data set with Talairach stereotaxic coordinate system.
The registration method is based on multi-scale, three-dimensional (3D) cross-correlation with an average (n > 300) MR brain image volume aligned with the Talariach stereotaxic space. Once the data set is re-sampled by the transformation recovered by the algorithm, atlas slices can be directly superimposed on the corresponding slices of the re-sampled volume. the use of such a standardized space also allows the direct comparison, voxel to voxel, of two or more data sets brought into stereotaxic space.
With use of a two-tailed Student t test for paired samples, there was no significant difference in the transformation parameters recovered by the automatic algorithm when compared with two manual landmark-based methods (p > 0.1 for all parameters except y-scale, where p > 0.05). Using root-mean-square difference between normalized voxel intensities as an unbiased measure of registration, we show that when estimated and averaged over 60 volumetric MR images in standard space, this measure was 30% lower for the automatic technique than the manual method, indicating better registrations. Likewise, the automatic method showed a 57% reduction in standard deviation, implying a more stable technique. The algorithm is able to recover the transformation even when data are missing from the top or bottom of the volume.
We present a fully automatic registration method to map volumetric data into stereotaxic space that yields results comparable with those of manually based techniques. The method requires no manual identification of points or contours and therefore does not suffer the drawbacks involved in user intervention such as reproducibility and interobserver variability.
The surface of the human cerebral cortex is a highly folded sheet with the majority of its surface area buried within folds. As such, it is a difficult domain for computational as well as visualization purposes. We have therefore designed a set of procedures for modifying the representation of the cortical surface to (i) inflate it so that activity buried inside sulci may be visualized, (ii) cut and flatten an entire hemisphere, and (iii) transform a hemisphere into a simple parameterizable surface such as a sphere for the purpose of establishing a surface-based coordinate system.
Several properties of the cerebral cortex, including its columnar and laminar organization, as well as the topographic organization of cortical areas, can only be properly understood in the context of the intrinsic two-dimensional structure of the cortical surface. In order to study such cortical properties in humans, it is necessary to obtain an accurate and explicit representation of the cortical surface in individual subjects. Here we describe a set of automated procedures for obtaining accurate reconstructions of the cortical surface, which have been applied to data from more than 100 subjects, requiring little or no manual intervention. Automated routines for unfolding and flattening the cortical surface are described in a companion paper. These procedures allow for the routine use of cortical surface-based analysis and visualization methods in functional brain imaging.
Clinical observation suggests that the aging process affects gyrification, with the brain appearing more 'atrophic' with increasing age. Empirical studies of tissue type indicate that gray matter volume decreases with age while cerebrospinal fluid increases. Quantitative changes in cortical surface characteristics such as sulcal and gyral shape have not been measured, however, due to difficulties in developing a method that separates abutting gyral crowns and opens up the sulci -- the 'problem of buried cortex'. We describe a quantitative method for measuring brain surface characteristics that is reliable and valid. This method is used to define the gyral and sulcal characteristics of atrophic and non-atrophic brains and to examine changes that occur with aging in a sample of 148 normal individuals from a broad age range. The shape of gyri and sulci change significantly over time, with the gyri becoming more sharply and steeply curved, while the sulci become more flattened and less curved. Cortical thickness also decreases over time. Cortical thinning progresses more rapidly in males than in females. The progression of these changes appears to be relatively stable during midlife and to begin to progress some time during the fourth decade. Measurements of sulcal and gyral shape may be useful in studying the mechanisms of both neurodevelopmental and neurodegenerative changes that occur during brain maturation and aging.
We present a novel, computerized method of examining cerebral cortical thickness. The normal cortex varies in thickness from 2 to 4 mm, reflecting the morphology of neuronal sublayers. Cortical pathologies often manifest abnormal variations in thickness, with examples of Alzheimer's disease and cortical dysplasia as thin and thick cortex, respectively. Radiologically, images are 2-D slices through a highly convoluted 3-D object. Depending on the relative orientation of the slices with respect to the object, it is impossible to deduce abnormal cortical thickness without additional information from neighboring slices. We approach the problem by applying Laplace's Equation (V2psi = 0) from mathematical physics. The volume of the cortex is represented as the domain for the solution of the differential equation, with separate boundary conditions at the gray-white junction and the gray-CSF junction. Normalized gradients of psi form a vector field, representing tangent vectors along field lines connecting both boundaries. We define the cortical thickness at any point in the cortex to be the pathlength along such lines. Key advantages of this method are that it is fully three-dimensional, and the thickness is uniquely defined for any point in the cortex. We present graphical results that map cortical thickness everywhere in a normal brain. Results show global variations in cortical thickness consistent with known neuroanatomy. The application of this technique to visualization of cortical thickness in brains with known pathology has broad clinical implications.
The finite mixture (FM) model is the most commonly used model for statistical segmentation of brain magnetic resonance (MR) images because of its simple mathematical form and the piecewise constant nature of ideal brain MR images. However, being a histogram-based model, the FM has an intrinsic limitation--no spatial information is taken into account. This causes the FM model to work only on well-defined images with low levels of noise; unfortunately, this is often not the the case due to artifacts such as partial volume effect and bias field distortion. Under these conditions, FM model-based methods produce unreliable results. In this paper, we propose a novel hidden Markov random field (HMRF) model, which is a stochastic process generated by a MRF whose state sequence cannot be observed directly but which can be indirectly estimated through observations. Mathematically, it can be shown that the FM model is a degenerate version of the HMRF model. The advantage of the HMRF model derives from the way in which the spatial information is encoded through the mutual influences of neighboring sites. Although MRF modeling has been employed in MR image segmentation by other researchers, most reported methods are limited to using MRF as a general prior in an FM model-based approach. To fit the HMRF model, an EM algorithm is used. We show that by incorporating both the HMRF model and the EM algorithm into a HMRF-EM framework, an accurate and robust segmentation can be achieved. More importantly, the HMRF-EM framework can easily be combined with other techniques. As an example, we show how the bias field correction algorithm of Guillemaud and Brady (1997) can be incorporated into this framework to achieve a three-dimensional fully automated approach for brain MR image segmentation.
We describe a sequence of low-level operations to isolate and classify brain tissue within T1-weighted magnetic resonance images (MRI). Our method first removes nonbrain tissue using a combination of anisotropic diffusion filtering, edge detection, and mathematical morphology. We compensate for image nonuniformities due to magnetic field inhomogeneities by fitting a tricubic B-spline gain field to local estimates of the image nonuniformity spaced throughout the MRI volume. The local estimates are computed by fitting a partial volume tissue measurement model to histograms of neighborhoods about each estimate point. The measurement model uses mean tissue intensity and noise variance values computed from the global image and a multiplicative bias parameter that is estimated for each region during the histogram fit. Voxels in the intensity-normalized image are then classified into six tissue types using a maximum a posteriori classifier. This classifier combines the partial volume tissue measurement model with a Gibbs prior that models the spatial properties of the brain. We validate each stage of our algorithm on real and phantom data. Using data from the 20 normal MRI brain data sets of the Internet Brain Segmentation Repository, our method achieved average kappa indices of kappa = 0.746 +/- 0.114 for gray matter (GM) and kappa = 0.798 +/- 0.089 for white matter (WM) compared to expert labeled data. Our method achieved average kappa indices kappa = 0.893 +/- 0.041 for GM and kappa = 0.928 +/- 0.039 for WM compared to the ground truth labeling on 12 volumes from the Montreal Neurological Institute's BrainWeb phantom.