Segmentation of Leukoaraiosis on Noncontrast Head CT Using CT‐MRI Paired Data Without Human Annotation

• Published in: Brain and behavior Vol. 15; no. 6; pp. e70602 - n/a
• Main Authors: Ryu, Wi‐Sun, Song, Jae W., Lim, Jae‐Sung, Lee, Ju Hyung, Sunwoo, Leonard, Kim, Dongmin, Kim, Dong‐Eog, Bae, Hee‐Joon, Lee, Myungjae, Kim, Beom Joon
• Format: Journal Article
• Language: English
• Published: United States John Wiley & Sons, Inc 01.06.2025; John Wiley and Sons Inc; Wiley
• Subjects: Aged; Aged, 80 and over; Algorithms; Automation; Brain cancer; computed tomography; Datasets; Deep Learning; Edema; Female; Humans; Ischemia; leukoaraiosis; Leukoaraiosis - diagnostic imaging; Magnetic fields; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Male; Medical imaging; Middle Aged; Neuroimaging; Neuroimaging - methods; Original; Republic of Korea; segmentation algorithm; Software; Stroke; Tomography, X-Ray Computed - methods; Traumatic brain injury; White Matter - diagnostic imaging; white matter hyperintensities; Republic of Korea; United States > US; computed tomography; deep learning; magnetic resonance imaging; segmentation algorithm; leukoaraiosis; white matter hyperintensities
• ISSN: 2162-3279; 2162-3279
• DOI: 10.1002/brb3.70602

ABSTRACT Objective Evaluating leukoaraiosis (LA) on CT is challenging due to its low contrast and similarity to parenchymal gliosis. We developed and validated a deep learning algorithm for LA segmentation using CT‐MRIFLAIR paired data from a multicenter Korean registry and tested it in a US dataset. Methods We constructed a large multicenter dataset of CT–FLAIR MRI pairs. Using validated software to segment white matter hyperintensity (WMH) on FLAIR, we generated pseudo‐ground‐truth LA labels on CT through deformable image registration. A 2D nnU‐Net architecture was trained solely on CT images and registered masks. Performance was evaluated using the Dice similarity coefficient (DSC), concordance correlation coefficient (CCC), and Pearson correlation across internal, external, and US validation cohorts. Clinical associations of predicted LA volume with age, risk factors, and poststroke outcomes were also analyzed. Results The external test set yielded a DSC of 0.527, with high volume correlations against registered LA (r = 0.953) and WMH (r = 0.951). In the external testing and US datasets, predicted LA volumes correlated with Fazekas grade (r = 0.832–0.891) and the correlations were consistent across CT vendors and infarct volumes. In an independent clinical cohort (n = 867), LA volume was independently associated with age, vascular risk factors, and 3‐month functional outcomes. Interpretation Our deep learning algorithm offers a reproducible method for LA segmentation on CT, bridging the gap between CT and MRI assessments in patients with ischemic stroke.