Lesion Explorer: A comprehensive segmentation and parcellation package to obtain regional volumetrics for subcortical hyperintensities and intracranial tissue

LC Campbell Cognitive Neurology Research Unit, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada.
NeuroImage (Impact Factor: 6.36). 01/2011; 54(2):963-73. DOI: 10.1016/j.neuroimage.2010.09.013
Source: PubMed


Subcortical hyperintensities (SH) are a commonly observed phenomenon on MRI of the aging brain (Kertesz et al., 1988). Conflicting behavioral, cognitive and pathological associations reported in the literature underline the need to develop an intracranial volumetric analysis technique to elucidate pathophysiological origins of SH in Alzheimer's disease (AD), vascular cognitive impairment (VCI) and normal aging (De Leeuw et al., 2001; Mayer and Kier, 1991; Pantoni and Garcia, 1997; Sachdev et al., 2008). The challenge is to develop processing tools that effectively and reliably quantify subcortical small vessel disease in the context of brain tissue compartments. Segmentation and brain region parcellation should account for SH subtypes which are often classified as: periventricular (pvSH) and deep white (dwSH), incidental white matter disease or lacunar infarcts and Virchow-Robin spaces. Lesion Explorer (LE) was developed as the final component of a comprehensive volumetric segmentation and parcellation image processing stream built upon previously published methods (Dade et al., 2004; Kovacevic et al., 2002). Inter-rater and inter-method reliability was accomplished both globally and regionally. Volumetric analysis showed high inter-rater reliability both globally (ICC=.99) and regionally (ICC=.98). Pixel-wise spatial congruence was also high (SI=.97). Whole brain pvSH volumes yielded high inter-rater reliability (ICC=.99). Volumetric analysis against an alternative kNN segmentation revealed high inter-method reliability (ICC=.97). Comparison with visual rating scales showed high significant correlations (ARWMC: r=.86; CHIPS: r=.87). The pipeline yields a comprehensive and reliable individualized volumetric profile for subcortical vasculopathy that includes regionalized (26 brain regions) measures for: GM, WM, sCSF, vCSF, lacunar and non-lacunar pvSH and dwSH.

24 Reads
  • Source
    • "is study demonstrates that WMH in periventricular and deep white matter are associated with decreased gray matter CBF and structural profiles in regions that are remote from the WMH lesions . The etiology of WMH remains a topic of intense research ( Thompson and Hakim , 2009 ; Debette and Markus , 2010 ; Gibson et al . , 2010 ; Uh et al . , 2010 ; Ramirez et al . , 2011 ; Makedonov et al . , 2013a ; van der Holst et al . , 2013 ; Wardlaw et al . , 2013 ) with one prevailing view that the lesions are caused by underlying vascular insufficiency ( Brickman et al . , 2009 ; Makedonov et al . , 2013b ; Wardlaw et al . , 2013 ) . A number of studies show that hypertension , diabetes , obesity and smoking are"
    [Show abstract] [Hide abstract]
    ABSTRACT: Cerebral White Matter Hyperintensities (WMH) are associated with vascular risk factors and age-related cognitive decline. WMH have primarily been associated with global white matter and gray matter (GM) changes and less is known about regional effects in GM. The purpose of this study was to test for an association between WMH and two GM imaging measures: cerebral blood flow (CBF) and voxel-based morphometry (VBM). Twenty-six elderly adults with mild to severe WMH participated in this cross-sectional 3 Tesla magnetic resonance imaging (MRI) study. MRI measures of GM CBF and VBM were derived from arterial spin labeling (ASL) and T1-weighted images, respectively. Fluid-attenuated inversion recovery (FLAIR) images were used to quantify the WMH lesion burden (mL). GM CBF and VBM data were used as dependent variables. WMH lesion burden, age and sex were used in a regression model. Visual rating of WMH with the Fazekas method was used to compare the WMH lesion volume regression approach. WMH volume was normally distributed for this group (mean volume of 22.7 mL, range: 2.2-70.6 mL). CBF analysis revealed negative associations between WMH volume and CBF in the left anterior putamen, subcallosal, accumbens, anterior caudate, orbital frontal, anterior insula, and frontal pole (corrected p < 0.05). VBM analysis revealed negative associations between WMH and GM volume in lingual gyrus, intracalcarine, and bilateral hippocampus (corrected p < 0.05). The visual rating scale corroborated the regression findings (corrected p < 0.05). WMH lesion volume was associated with intra-group GM CBF and structural differences in this cohort of WMH adults with mild to severe lesion burden.
    Frontiers in Aging Neuroscience 07/2015; 7:131. DOI:10.3389/fnagi.2015.00131 · 4.00 Impact Factor
  • Source
    • "These landmarks and tracings are then automatically combined with the Talaraich-like grid to generate a parcellation mask (Fig. 1A, B and H) that can be applied to the tissue segmentation to generate 26 regions of interest (13 per hemisphere ) for each of the tissue classes (Dade et al., 2004; Ramirez et al., 2011). Further segmentation of white matter hyperintensity (WMH) tissue classes is achieved through a semi-automated technique called " Lesion Explorer, " which utilizes a tri-feature algorithm based on PD/T2 and T1 to quantify hyperintensities into 4 sub-classes (Fig. 1I and J) including both periventricular and deep hyperintensities and CSF-filled " black holes " , with minor manual editing by a highly trained rater to remove false positives (Ramirez et al., 2011). Together these pipelines yield individualized subjectspecific volumetric data for 8 tissue classes (Fig. 1K) for up to 26 regions of interest, which can be further combined to create summary volumetrics for various total lobes (medial/lateral frontal, total parietal, temporal, etc.). "
    [Show abstract] [Hide abstract]
    ABSTRACT: We report a single-case study of a female patient (VL) who exhibited frequent episodes of erroneous recollections triggered by everyday events. Based on neuropsychological testing, VL was classified as suffering from mild to moderate dementia (MMSE=18) and was given a diagnosis of probable Alzheimer’s disease. Her memory functions were uniformly impaired but her verbal abilities were generally well preserved. A structural MRI scan showed extensive areas of gray matter atrophy particularly in frontal and medial-temporal (MTL) areas. Results of experimental recognition tests showed that VL had very high false alarm rates on tests using pictures, faces and auditory stimuli, but lower false alarm rates on verbal tests. We provide a speculative account of her erroneous recollections in terms of her MTL and frontal pathology. In outline, we suggest that owing to binding failures in MTL regions, VL’s recognition processes were forced to rely on earlier than normal stages of analysis. Environmental features on a given recognition trial may have combined with fragments persisting from previous trials resulting in erroneous feelings of familiarity and of recollection that were not discounted or edited out, due to her impaired frontal processes.
    Neuropsychologia 04/2014; 56(1). DOI:10.1016/j.neuropsychologia.2014.02.007 · 3.30 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Treatment for medulloblastoma during childhood impairs neurocognitive function in survivors. While those diagnosed at younger ages are most vulnerable, little is known about the long-term neurocognitive, functional, and physical outcomes in survivors as they approach middle age. In this retrospective cohort study, we assessed 20 adults who were treated with surgery and radiotherapy for medulloblastoma during childhood (median age at assessment, 21.9 years [range, 18-47 years]; median time since diagnosis, 15.5 years [range, 6.5-42.2 years]). Nine patients also underwent chemotherapy. Cross-sectional analyses of current neurocognitive, functional, and physical status were conducted. Data from prior neuropsychological assessments were available for 18 subjects; longitudinal analyses were used to model individual change over time for those subjects. The group was well below average across multiple neurocognitive domains, and 90% had required accommodations at school for learning disorders. Longer time since diagnosis, but not age at diagnosis, was associated with continued decline in working memory, a common sign of aging. Younger age at diagnosis was associated with lower intelligence quotient and academic achievement scores, even many years after treatment had been completed. The most common health complications in survivors were hearing impairment, second cancers, diabetes, hypertension, and endocrine deficiencies. Adult survivors of childhood medulloblastoma exhibit signs of early aging regardless of how young they were at diagnosis. As survival rates for brain tumors continue to improve, these neurocognitive and physical sequelae may become evident in survivors diagnosed at different ages across the lifespan. It will become increasingly important to identify factors that contribute to risk and resilience in this growing population.
    Neuro-Oncology 03/2011; 13(5):536-45. DOI:10.1093/neuonc/nor015 · 5.56 Impact Factor
Show more

Similar Publications