Meeting Banner
Abstract #1098

Automated Detection, Segmentation, and Longitudinal Tracking of Active MS Lesions Via Subtraction MRI

Colin D. Shea1, Navid Shiee2, Emily Wood1, Dzung Pham2, Govind Bhagavatheeshwaran1, Daniel S. Reich1

1NINDS, National Institutes of Health, Bethesda, MD, United States; 2Diagnostic Radiology, National Institutes of Health, Bethesda, MD, United States

We present an automated method based on subtraction MRI to detect, delineate, and track new and changing MS lesions in order to study the spatiotemporal dynamics of lesions in large datasets.

Keywords

active activity agreement allowing analyzed appearing arrows artifacts assessment automated binary biological blue bottom brain called changing chronological classification combined consecutive consistent context contrast dataset datasets days delineate delineated density derived detail detect detecting detection developed diagnostic disease distinguished duration dynamics efficacy enables enhanced enhancing evolving existing expanding expansion fashion flair generate geometric green health identified improves indicate individual individuals input institutes intensity interval joining least lesion lesions load longitudinal maintain manually mask masks measures median medical monitoring national newly noise normalized object overlapped pairwise patients predictive preserved previous previously procedure processes processing progression propose proton radiology randomly recent registration robust sample scale scanner scheme science sclerosis segmentation segmentations segmented segmenting selected sensitive sensitivity sentinel shrinking specificity studies subjects subtracting subtraction third thresholding tissue tool toolkit track tracking treatment trials unfeasible validate visual visualization visually volumes white wood years