Search

The detection of contrast-enhancing lesions (CELs) is fundamental for the diagnosis and monitoring of patients with multiple sclerosis (MS). This task is time-consuming and suffers from high intra- and inter-rater variability in clinical practice. However, only a few studies proposed automatic approaches for CEL detection. This study aimed to develop a deep learning model that automatically detects and segments CELs in clinical Magnetic Resonance Imaging (MRI) scans. A 3D UNet-based network was trained with clinical MRI from the Swiss Multiple Sclerosis Cohort. The dataset comprised 372 scans from 280 MS patients: 162 showed at least one CEL, while 118 showed no CELs. The input dataset consisted of T1-weighted before and after gadolinium injection, and FLuid Attenuated Inversion Recovery images. The sampling strategy was based on a white matter lesion mask to confirm the existence of real contrast-enhancing lesions. To overcome the dataset imbalance, a weighted loss function was implemented. The Dice Score Coefficient and True Positive and False Positive Rates were 0.76, 0.93, and 0.02, respectively. Based on these results, the model developed in this study might well be considered for clinical decision support. [ABSTRACT FROM AUTHOR]

Objective Methods Results Interpretation Pathological studies suggest that multiple sclerosis (MS) lesions endure multiple waves of damage and repair; however, the dynamics and characteristics of these processes are poorly understood in patients living with MS.We studied 128 MS patients (75 relapsing–remitting, 53 progressive) and 72 healthy controls who underwent advanced magnetic resonance imaging and clinical examination at baseline and 2 years later. Magnetization transfer saturation and multi‐shell diffusion imaging were used to quantify longitudinal changes in myelin and axon volumes within MS lesions. Lesions were grouped into 4 classes (repair, damage, mixed repair damage, and stable). The frequency of each class was correlated to clinical measures, demographic characteristics, and levels of serum neurofilament light chain (sNfL).Stable lesions were the most frequent (n = 2,276; 44%), followed by lesions with patterns of “repair” (n = 1,352; 26.2%) and damage (n = 1,214; 23.5%). The frequency of “repair” lesion was negatively associated with disability (β = −0.04; p < 0.001) and sNfL (β = −0.16; p < 0.001) at follow‐up. The frequency of the “damage” class was higher in progressive than relapsing–remitting patients (p < 0.05) and was related to disability (baseline: β = −0.078; follow‐up: β = −0.076; p < 0.001) and age (baseline: β = −0.078; p < 0.001). Stable lesions were more frequent in relapsing–remitting than in progressive patients (p < 0.05), and in younger patients versus older (β = −0.07; p < 0.001) at baseline. Further, “mixed” lesions were most frequent in older patients (β = 0.004; p < 0.001) at baseline.These findings show that repair and damage processes within MS lesions occur across the entire disease spectrum and that their frequency correlates with patients disability, age, disease duration, and extent of neuroaxonal damage. ANN NEUROL 2024 [ABSTRACT FROM AUTHOR]