Subject‐Level Segmentation Precision Weights for Volumetric Studies Involving Label Fusion

• Published in: Human brain mapping Vol. 45; no. 18; pp. e70082 - n/a
• Main Authors: Chen, Christina, Das, Sandhitsu R., Tisdall, M. Dylan, Hu, Fengling, Chen, Andrew A., Yushkevich, Paul A., Wolk, David A., Shinohara, Russell T.
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 15.12.2024
• Subjects: Aged; Aged, 80 and over; Algorithms; Alzheimer Disease - diagnostic imaging; Alzheimer Disease - pathology; Alzheimer's disease; Automation; Brain - diagnostic imaging; Cognitive ability; Cognitive Dysfunction - diagnostic imaging; Cognitive Dysfunction - pathology; Cognitive Dysfunction - physiopathology; Disease; Disease control; Estimates; Female; Females; Hippocampus; Hippocampus - diagnostic imaging; Humans; Image processing; Image Processing, Computer-Assisted - methods; Image Processing, Computer-Assisted - standards; Image segmentation; Information processing; Labels; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Medical imaging; Middle Aged; Neurodegenerative diseases; Neuroimaging; Neuroimaging - methods; Neuroimaging - standards; Organ Size - physiology; Phenotypes; Phenotypic variations; Registration; Segmentation
• ISSN: 1065-9471; 1097-0193; 1097-0193
• DOI: 10.1002/hbm.70082

ABSTRACT In neuroimaging research, volumetric data contribute valuable information for understanding brain changes during both healthy aging 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 naïve approach hinders comparing ROI volumes between different samples to identify associations between tissue volume and disease or phenotype. We propose a novel method that estimates the variance of the measured ROI volume for each subject due to the multi‐atlas segmentation procedure. We demonstrate 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. We propose a novel method that estimates subject‐level precision weights for ROI volume estimates produced by multi‐atlas segmentation. We demonstrate in real data that incorporating these weights markedly improves the power to detect a mean difference in hippocampal volume between controls and subjects with mild cognitive impairment or Alzheimer's disease.