December 2024
·
3 Reads
This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.
December 2024
·
3 Reads
December 2024
·
31 Reads
The human brain is a complex and highly dynamic system, and our current knowledge of its functional mechanism is still very limited. Fortunately, with functional magnetic resonance imaging (fMRI), we can observe blood oxygen level-dependent (BOLD) changes, reflecting neural activity, to infer brain states and dynamics. In this paper, we ask the question of whether the brain states rep-resented by the regional brain fMRI can be predicted. Due to the success of self-attention and the transformer architecture in sequential auto-regression problems (e.g., language modelling or music generation), we explore the possi-bility of the use of transformers to predict human brain resting states based on the large-scale high-quality fMRI data from the human connectome project (HCP). Current results have shown that our model can accurately predict the brain states up to 5.04s with the previous 21.6s. Furthermore, even though the prediction error accumulates for the prediction of a longer time period, the gen-erated fMRI brain states reflect the architecture of functional connectome. These promising initial results demonstrate the possibility of developing gen-erative models for fMRI data using self-attention that learns the functional or-ganization of the human brain. Our code is available at: https://github.com/syf0122/brain_state_pred.
September 2024
·
32 Reads
Diffusion-weighted imaging (DWI) is a type of Magnetic Resonance Imaging (MRI) technique sensitised to the diffusivity of water molecules, offering the capability to inspect tissue microstructures and is the only in-vivo method to reconstruct white matter fiber tracts non-invasively. The DWI signal can be analysed with the diffusion tensor imaging (DTI) model to estimate the directionality of water diffusion within voxels. Several scalar metrics, including axial diffusivity (AD), mean diffusivity (MD), radial diffusivity (RD), and fractional anisotropy (FA), can be further derived from DTI to quantitatively summarise the microstructural integrity of brain tissue. These scalar metrics have played an important role in understanding the organisation and health of brain tissue at a microscopic level in clinical studies. However, reliable DTI metrics rely on DWI acquisitions with high gradient directions, which often go beyond the commonly used clinical protocols. To enhance the utility of clinically acquired DWI and save scanning time for robust DTI analysis, this work proposes DirGeo-DTI, a deep learning-based method to estimate reliable DTI metrics even from a set of DWIs acquired with the minimum theoretical number (6) of gradient directions. DirGeo-DTI leverages directional encoding and geometric constraints to facilitate the training process. Two public DWI datasets were used for evaluation, demonstrating the effectiveness of the proposed method. Extensive experimental results show that the proposed method achieves the best performance compared to existing DTI enhancement methods and potentially reveals further clinical insights with routine clinical DWI scans.
April 2024
·
251 Reads
Human subcortical grey matter (SGM) structures are crucial hubs for processing, transmitting and modulating brain-wide signals. However, in-vivo characterisation of subnuclei within the SGM structures remains limited, restricting our knowledge of their roles in brain function and in disorders. To address this gap, we introduce the Sydney Subcortical Gray Matter (SydSGM) parcellation— based on a novel data-driven approach using multiple MRI contrasts to assess myelination, directionality of white matter fibres, and diffusion micro-environment. Fifty-four distinct parcels were identified across caudate, putamen, globus pallidus, nucleus accumbens, and thalamus based on the coherent representation of these attributes, which demonstrated remarkable concordance with histology-based delineations. The SydSGM parcellation now facilitates detailed structural connectivity analysis at the subnuclei scale. It can be adopted as a standalone atlas or incorporated into existing brain atlases. We demonstrate SydSGM parcellation’s advantage by revealing declined structural connectivity of highly resolved SGM subnuclei in patients with early Parkinson’s disease.
April 2024
·
11 Reads
·
3 Citations
Artificial Intelligence in Medicine
March 2024
·
60 Reads
·
3 Citations
In medicine, abnormalities in quantitative metrics such as the volume reduction of one brain region of an individual versus a control group are often provided as deviations from so-called normal values. These normative reference values are traditionally calculated based on the quantitative values from a control group, which can be adjusted for relevant clinical co-variables, such as age or sex. However, these average normative values do not take into account the globality of the available quantitative information. For example, quantitative analysis of T1-weighted magnetic resonance images based on anatomical structure segmentation frequently includes over 100 cerebral structures in the quantitative reports, and these tend to be analyzed separately. In this study, we propose a global approach to personalized normative values for each brain structure using an unsupervised Artificial Intelligence technique known as generative manifold learning. We test the potential benefit of these personalized normative values in comparison with the more traditional average normative values on a population of patients with drug-resistant epilepsy operated for focal cortical dysplasia, as well as on a supplementary healthy group and on patients with Alzheimer’s disease.
March 2024
·
42 Reads
·
1 Citation
This study introduces a novel brain connectome matrix, track‐weighted PET connectivity ( twPC ) matrix, which combines positron emission tomography ( PET ) and diffusion magnetic resonance imaging data to compute a PET ‐weighted connectome at the individual subject level. The new method is applied to characterise connectivity changes in the Alzheimer's disease (AD) continuum. The proposed twPC samples PET tracer uptake guided by the underlying white matter fibre‐tracking streamline point‐to‐point connectivity calculated from diffusion MRI ( dMRI ) . Using tau‐PET , dMRI and T1 ‐weighted MRI from the Alzheimer's Disease Neuroimaging Initiative database, structural connectivity ( SC ) and twPC matrices were computed and analysed using the network‐based statistic (NBS) technique to examine topological alterations in early mild cognitive impairment ( MCI ), late MCI and AD participants. Correlation analysis was also performed to explore the coupling between SC and twPC . The NBS analysis revealed progressive topological alterations in both SC and twPC as cognitive decline progressed along the continuum. Compared to healthy controls, networks with decreased SC were identified in late MCI and AD , and networks with increased twPC were identified in early MCI , late MCI and AD . The altered network topologies were mostly different between twPC and SC , although with several common edges largely involving the bilateral hippocampus, fusiform gyrus and entorhinal cortex. Negative correlations were observed between twPC and SC across all subject groups, although displaying an overall reduction in the strength of anti‐correlation with disease progression. twPC provides a new means for analysing subject‐specific PET and MRI ‐derived information within a hybrid connectome using established network analysis methods, providing valuable insights into the relationship between structural connections and molecular distributions. Practitioner Points New method is proposed to compute patient‐specific PET connectome guided by MRI fibre‐tracking. Track‐weighted PET connectivity (twPC) matrix allows to leverage PET and structural connectivity information. twPC was applied to dementia, to characterise the PET nework abnormalities in Alzheimer's disease and mild cognitive impairment.
February 2024
·
8 Reads
Objectives Structural connectivity and graph theory analysis may elucidate substrates for cognitive impairment in MS, providing additional biomarkers for disease progression in addition to conventional MRI analysis. We investigated the association between global network metrics and cognitive impairment cross-sectionally and longitudinally over 24 months. Methods 48 patients with relapsing-remitting MS were recruited for an observational study over 24 months. Processing speed, learning/memory, executive function and language were assessed with the Minimal Assessment of Cognitive Function in MS (MACFIMS) battery. Patients were divided into cognitively-impaired (Z score <-1SD for ≥ 2 tests) versus cognitively-preserved at baseline. Structural connectomes were reconstructed using diffusion-weighted imaging and probabilistic tractography.Network metrics and cognitive scores were analysed at baseline, and network metric changes were correlated with cognitive changes over 24 months. Results Mean age was 30 years, 63% female, median disease duration 1.88 years, median EDSS 2.0. At baseline, network metrics were not correlated with any cognitive outcomes after adjusting for age, gender and T2 lesion volume. Between cognitively impaired (n=25) and preserved patients (n=23), there was no difference in lesion volume, normalised gray matter volume, or network metrics. Baseline EDSS was not significantly correlated with any graph measures. Over two years, change in executive function was negatively correlated with change in mean local efficiency, clustering co- efficient and assortativity (p<0.05). Conclusion Global network metrics were not correlated with cognitive outcomes or EDSS scores, and did not differentiate between cognitively impaired vs preserved patients. Longitudinal changes in global network metrics are associated with cognitive decline.
December 2023
·
50 Reads
Background The three clinical presentations of frontotemporal dementia (FTD), behavioral‐variant FTD (bvFTD), progressive non‐fluent aphasia (PNFA), and semantic dementia (SD), are caused by abnormal aggregation of protein pathologies and progressive brain atrophy that precede the emergence of clinical symptoms. Neuropathology shows a ‘prion‐like’ propagation along white matter pathways before aggregating in the neuronal cell bodies of the grey matter. The in‐vivo mapping of upstream white matter trajectories of degeneration can improve disease staging and understanding of disease mechanisms in FTD, but longitudinal white matter studies are rare due to methodological challenges. Here, we develop a novel neuroimaging approach to map white matter changes in FTD using fixel‐based analysis (FBA), a new diffusion weighted imaging technique. Method Sixty FTD patients (23 bvFTD, 15 PNFA and 22 SD) were matched with 30 heathy controls. Participants underwent a comprehensive annual clinical assessment and high‐resolution multimodal MRI (median 2 years; range 1‐6 years). Time‐varying changes in white matter fibre density and cross‐section were computed with MRtrix3, FreeSurfer 7.1.1 and Spatiotemporal Linear Mixed Effects models. Result All FTD syndromes showed more extensive white matter changes at baseline than previously reported. In bvFTD, atrophy extended posteriorly over time, encroaching in tracts connecting subcortical and motor‐associations regions (Fig 1). In PNFA, baseline left‐lateralized atrophy extended anteriorly and inferiorly and into the contralateral hemisphere, eventually mirroring patterns of atrophy at presentation (Fig 2). In SD, the focal left temporal baseline atrophy extended posteriorly and laterally along the inferior and superior longitudinal fasculi, and involving the right hemisphere with disease progression (Fig 3). Conclusion The FBA approach enabled the identification of new white matter bundles in FTD, with syndrome‐specific effects and improved biological interpretability compared to traditional tensor imaging methods. Our method can be extended beyond FTD to other neurodegenerative disorders to inform better models of disease staging and provide useful targets for patient stratification and monitoring in trials of disease‐monitoring interventions.
September 2023
·
74 Reads
·
7 Citations
Frontiers in Radiology
Diffusion-weighted Imaging (DWI) is a non-invasive imaging technique based on Magnetic Resonance Imaging (MRI) principles to measure water diffusivity and reveal details of the underlying brain micro-structure. By fitting a tensor model to quantify the directionality of water diffusion a Diffusion Tensor Image (DTI) can be derived and scalar measures, such as fractional anisotropy (FA), can then be estimated from the DTI to summarise quantitative microstructural information for clinical studies. In particular, FA has been shown to be a useful research metric to identify tissue abnormalities in neurological disease (e.g. decreased anisotropy as a proxy for tissue damage). However, time constraints in clinical practice lead to low angular resolution diffusion imaging (LARDI) acquisitions that can cause inaccurate FA value estimates when compared to those generated from high angular resolution diffusion imaging (HARDI) acquisitions. In this work, we propose High Angular DTI Estimation Network (HADTI-Net) to estimate an enhanced DTI model from LARDI with a set of minimal and evenly distributed diffusion gradient directions. Extensive experiments have been conducted to show the reliability and generalisation of HADTI-Net to generate high angular DTI estimation from any minimal evenly distributed diffusion gradient directions and to explore the feasibility of applying a data-driven method for this task. The code repository of this work and other related works can be found at https://mri-synthesis.github.io/ .
... The convolutional and pooling layers gather the high-level features to upsample them to the resolution of the initial image using a specific type of tied unpooling and convolutional layers. In [35] explore federated learning to enhance MS lesion segmentation in MRI, addressing data privacy and label noise issues. Introducing DHLC and CELC, our methods improve annotation accuracy and model robustness. ...
April 2024
Artificial Intelligence in Medicine
... The selection of similar patients may be distance-based according to a patient-similarity metric. This may imply a digital representation of patients as data points in a latent reduced multidimensional space, using linear or nonlinear dimension-reduction algorithms [60,61]. On the other hand, the selection of similar patients may be filter-based. ...
March 2024
... This approach utilizes limited dMRI volumes to generate reliable diffusion metrics comparable to those derived from sufficient dMRI volumes. The feasibility of angular resolution enhancement has been explored in recent studies [17,3,5]. In the training process, the generator G A accepts artifact-free b 0 image concatenated with 6 unique DWIs (b 0 ⊕ 6DW Is) of a subject as inputs X ∈ R N ×7×W ×H×D and transforms X into an angular resolution enhanced FA map, denoted as F A ∈ R N ×1×W ×H×D . ...
September 2023
Frontiers in Radiology
... BrEd in FSL is a visualization tool developed for inspecting NIfTI images with different Lightwave buttons for image manipulation like zoom, window, and sending to the overlay manager. [49][50][51] FSL and Python libraries also interface with Python. This allows execution of FSL on a map with a script without opening the FSL graphical user interface for projects. ...
January 2023
... The segmentation findings reached a DSC=72%, outperforming the performance of U-net alone and U-net with the normal attention unit. Comparing the results with another study [47] that used the same public dataset including also data from three different hospitals, achieved a DSC=62%. ...
May 2023
... The copyright holder for this this version posted December 11, 2024. ; https://doi.org/10.1101/2024.12.06.24318287 doi: medRxiv preprint 2019; Kapadia et al., 2020;Purrer et al., 2021;Kyle et al., 2023). Similarly, we observed that after one month overall lesion volume marginally correlated with concurrent improvement in tremor severity, but it was not associated with tremor change after six months. ...
April 2023
... 42 To date, most neuroimaging studies have used group-based analyses that are unable to account for the significant individual differences in mTBI presentations. 43 Applying the concept of clinical profiles or subtypes to MRI analysis may provide more meaningful insight relating microstructural injury in brain regions specific to an individual's clinical presentation. 20,42,44 Improving our understanding of the cause of post-mTBI symptoms provides valuable insight into possible management approaches. ...
February 2023
Brain Research
... Consisting of more than 50% of the total neurons in the whole brain ( Wagner et al., 2019;Zagon et al., 1977), the cerebellum has a repeated, modular structure that integrates the cerebellar cortex with the cerebral cortex, both via inputs (via the cortico-pontine-cerebellar system) and outputs (via the cerebello-thalamo-cortical circuits) ( Albus, 1971;Koziol et al., 2014;Nashef et al., 2018), as well as with the spinal cord ( Albus, 1971;Cohen et al., 2017;Zinger et al., 2013). The modular structure of the cerebellum allows for compartmentalisation of unique inputs ( Hayter et al., 2007;Kostadinov & Häusser, 2022), the learning and execution of complex, nested sequences of input-output patterns ( Khilkevich et al., 2018), as well as more complex, multidomain functions like parallel processing ( Müller et al., 2023). Furthermore, the cerebellum has also been linked to pattern separation and feed-forward predictions, akin to a forward model for effective motor control ( Bastian, 2006;Imamizu & Kawato, 2009;Ramnani, 2014;Stein, 2021;Wolpert et al., 1998). ...
June 2023
... Two groups used two different divisions of the three subsets, using the first one to determine the best method, and the Table 1 Employed AI-reconstruction methods to synthesize FA, AD and MD from 21 gradient directions. FCN: Fully Connected Network; CNN: Convolutional neural network; MLP: Multilayer perceptron; UN: Unrolled network (Ye et al., 2020;HashemizadehKolowri et al., 2022); U-Net (Çiçek et al., 2016;Tang et al., 2021, Tang et al., 2022; DAE: Denoising autoencoder (Faiyaz et al., 2022b); AEME: Adaptive network with extragradient for dMRI based microstructure estimation (Zheng et al., 2022); SARDI-Net: Super-angular Resolution Diffusion Imaging Network ; DNSR: Dual network scoring and reconstruction (Blumberg et al., 2022); Swin-Transformer . The term enhanced metrics used in the description of the methods refers to the 3 considered metrics (FA, MA, AD) calculated by the different methods to achieve a quality similar to the parameters estimated from 61 gradient directions. ...
December 2022
Lecture Notes in Computer Science
... Decreased functional connectivity in the DMN has been reported in many studies . T his pattern of DMN dysfunction has been displayed in the MCI group with limited increases compared with the CN group between DMN structures, indicative of a pr odr omal compensatory mec hanism (DeMayo et al., 2023a ;Niu et al., 2018 ;Zhang et al., 2020 ;Zheng et al., 2017 ). Brain structures implicated in the DMN mostly include the posterior cingulate cortex (PCC), pr ecuneus cortex, medial pr efr ontal and later al tempor al cortices and hippocampus (Forouzannezhad et al., 2019 ). ...
November 2022
Brain Connectivity