A0583
Title: Subject-level segmentation accuracy weights for volumetric studies involving label fusion
Authors: Christina Chen - University of Pennsylvania (United States) [presenting]
Sandhitsu Das - University of Pennsylvania (United States)
Matthew Tisdall - University of Pennsylvania (United States)
Fengling Hu - University of Pennsylvania (United States)
Andrew Chen - University of Pennsylvania (United States)
Paul Yushkevich - University of Pennsylvania (United States)
David Wolk - University of Pennsylvania (United States)
Russell Shinohara - University of Pennsylvania (United States)
Abstract: In neuroimaging research, volumetric data contribute valuable information for understanding brain changes during healthy ageing and pathological processes. Extracting these measures from images requires segmenting the regions of interest (ROIs), and many popular methods accomplish this by fusing labels from multiple expert-segmented images called atlases. However, post-segmentation, current practices typically treat each subject's measurement equally without incorporating any information about variation in their segmentation precision. This naive approach hinders comparing ROI volumes between different samples to identify associations between tissue volume and disease or phenotype. A novel method is proposed that estimates the variance of the measured ROI volume for each subject due to the multi-atlas segmentation procedure. It is demonstrated in real data that weighting by these estimates markedly improves the power to detect a mean difference in hippocampal volume between controls and subjects with mild cognitive impairment or Alzheimer's disease.
