Generalizable spinal cord multiple sclerosis lesion segmentation across MRI contrasts, protocols, and centers

Key Points

• This research aims to improve spinal cord multiple sclerosis lesion segmentation using a reliable model across different MRI data.
• Model implemented in Spinal Cord Toolbox v7.2 and higher.
• Segmentation tested on heterogeneous MRI data from multiple centers.
• Focus on accuracy and reliability of model in clinical settings.
• Achieved accurate segmentation of MS lesions across various MRI protocols.
• Model effectively addresses barriers for clinical translation of MRI analysis.
• Demonstrated reliability across different data types and centers.