Poster No:
2124
Submission Type:
Abstract Submission
Authors:
Sheena Baratono1, Grace Burt2, William Drew2, Michael Fox3
Institutions:
1Brigham and Women's Hospital, Brookline, MA, 2Brigham and Women's Hospital, Boston, MA, 3Brigham and Women’s Hospital, Harvard Medical School, Boston, MA
First Author:
Co-Author(s):
Grace Burt
Brigham and Women's Hospital
Boston, MA
Michael Fox
Brigham and Women’s Hospital, Harvard Medical School
Boston, MA
Introduction:
Several MRI methods have been used to define atrophy in neurodegenerative diseases including Alzheimer's disease. Raw cortical thickness measures are commonly compared across groups but are confounded by regional heterogeneity due to intrinsic differences in gyral/sulcal thickness(1). To account for this heterogeneity and allow comparisons across cohorts, people have started using w-maps. A w-map is comprised of z-scores at every vertex, comparing that individual's cortical thickness against an expected cortical thickness based on a normative group. However, it is unclear whether w-maps should be thresholded and focused on peak atrophy to exclude noise or if the more subtle variations in relative cortical thickness provide pertinent information.
Methods:
In this study, cortical thickness and multiple w-mapping methods were compared using the Alzheimer's Disease Neuroimaging Initiative cohorts (ADNI1 = 183 AD, 215 controls (CN), ADNI2 = 140 AD and 188 CN). MPRAGE MRIs sequences were processed in surface space using Freesurfer v7.2. Cortical thickness is determined on a vertex wise basis comparing the distance between the pial and white mater surfaces. These raw cortical thickness files were converted to GIFTIs and used for cortical thickness analyses. The w-maps were generated using cohort specific general linear models (GLMs) created from controls to calculate z scores while covarying for age and gender. Control subjects were excluded from their own GLM. A negative z-score occurs when the observed cortical thickness is thinner than expected by the GLM divided by the standard deviation. Unthresholded w-maps were then compared to w-maps thresholded at a z-score of negative 2 to define peak atrophy as has been done in the literature(2).
Methods were compared at the group and single subject levels. The w-maps were overlapped to create group level AD and CN maps for ADNI1 and for ADNI2. A two-sample t-test was performed comparing AD single subject w-maps to their Alzheimer's Disease Assessment Scale - cognitive sub-scale (ADAS-cog) 11 scores for each cohort. The ADAS-cog 11 is a normalized test of general cognition. Reproducibility was assessed using the ground truth spatial correlation comparing ADNI1 to ADNI2.
Results:
Cortical thickness analyses demonstrate sulcal/gyral differences in both AD and CNs which is not seen in unthresholded w-maps. Group level overlap maps for AD subjects for both unthresholded and z= -2 thresholded w-maps mimic known atrophy patterns in AD and highly correlate with group level AD vs CN two sample t-test maps. However, the CN group level map created from the -2 thresholded maps also resembles the AD atrophy pattern, while group level unthresholded maps do not show a clear atrophy spatial bias and do not correlate with the AD v CN map. Reproducibility between ADNI1 and ADNI2 ADAS-cog 11 was limited but higher in the unthresholded (0.22, p<0.004) compared to -2 thresholded w-maps (0.16, p<0.002).
Conclusions:
In addition to normalizing cohorts, w-mapping removes regional heterogeneity seen in cortical thickness. Using unthresholded vs thresholded w-maps leads to better reproducibility and eliminates any spatial bias of atrophy within controls. These findings demonstrate the utility of w-mapping when assessing atrophy and show that the subtle variations in relative cortical thickness outside of peak regions positively impacts reproducibility.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Higher Cognitive Functions:
Higher Cognitive Functions Other
Modeling and Analysis Methods:
Image Registration and Computational Anatomy
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping 1
Novel Imaging Acquisition Methods:
Anatomical MRI
Keywords:
Aging
Cognition
Degenerative Disease
Spatial Normalization
Statistical Methods
Other - atrophy
1|2Indicates the priority used for review
Provide references using author date format
(1) Salat, D.H. et al (2004), 'Thinning of the Cerebral Cortex in Aging'. Cerebral Cortex, V14 N7 pp 721-730
(2) Tetreault, A.M. et al (2020), 'Network Localization of Clinical, Cognitive and Neuropsychiatric Symptoms in Alzheimer's Disease.' BRAIN, V143 pp1249-1260