T1-Dark Rim Lesion Training Dataset: Expert-Labeled MRI Cases for Standardized Identification in Multiple Sclerosis

This dataset provides expert-labeled MRI cases for training and evaluating identification of T1-dark rim lesions (T1-DRLs), a recently described imaging sign of chronic active inflammation in multiple sclerosis. Unlike paramagnetic rim lesions requiring specialized susceptibility-weighted imaging, T1-DRLs are detectable on routine 3D T1-weighted sequences universally acquired across centers and vendors, enabling assessment in large retrospective cohorts and routine clinical practice. Dataset contents: 396 expert-labeled lesion cases (246 true T1-DRLs, 150 false positives) 7,925 anonymized MRI images in WEBP format (11–49 consecutive slices per lesion) 20 interactive batches: 5 guided training batches with visible labels, 15 self-evaluation batches with hidden labels Consensus labels and metadata in CSV format Fully functional web-based training platform (HTML/CSS/JavaScript) Labeling methodology: Labels were established through a two-stage consensus process: (1) sensitivity-oriented candidate marking by two trained readers, followed by (2) specificity-oriented expert consensus determination by a neuroradiologist and senior resident with extensive multiple sclerosis imaging experience, applying published T1-DRL criteria. Intended uses: Self-directed training for radiologists and neurologists before research participation or clinical adoption Standardized reader training for multi-center validation studies Ground truth data for developing automated detection algorithms Inter-rater reliability studies with pre-defined consensus labels Integration into neuroradiology educational curricula The web platform requires no software installation, functions on any device with a modern browser, and stores user progress locally. All images are fully anonymized and approved for educational sharing (Ethics Committee approval PR268/24). Citation: 1. Naval-Baudin P, Pons-Escoda A, Castillo-Pinar A, et al. The T1-dark-rim: A novel imaging sign for detecting smoldering inflammation in multiple sclerosis. Eur J Radiol 2024;173:111358. https://doi.org/10.1016/j.ejrad.2024.111358