Comparing Lesion Volume Dynamics Between Multiple Sclerosis and Neuromyelitis Optica Spectrum Disorder During Remission Using Machine-Learning Segmentation.

background and purposeMultiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD) are inflammatory demyelinating conditions of the central nervous system that have distinct pathological mechanisms. There is a paucity of studies comparing the accumulation of subclinical lesions between MS and NMOSD, especially during the clinical remission period. Recent advances in neuroimaging techniques, those particularly involving the use of machine learning (ML) methods for lesion segmentation, have provided new opportunities to quantitatively assess the volumes of brain lesions and how they change over time. In this study, we aimed to use ML-based lesion segmentation methods to measure differences in lesion volumes and their changes during the remission period between patients with MS and NMOSD.

methodsThis study included a retrospective cohort of 31 patients with MS and patients with 30 aquaporin-4-positive (AQP4⁺) NMOSD from the National Cancer Center registry. Serial 3D brain magnetic resonance imaging (MRI) scans obtained during the interattack period were analyzed using ML-based segmentation. MRI data preprocessing included alignment, distortion correction, and normalization, with lesion mapping and statistical analyses determining changes in lesion volumes.

resultsThe MS patients exhibited significant increases in the median lesion volume (from 3,493 mm³ to 4,430 mm³,

conclusionsThese findings highlight the distinct pathogenic processes of the two conditions and hence the need for specialized therapeutic and monitoring strategies for patients with MS and AQP4⁺ NMOSD.