Brain tissue segmentation in multiple sclerosis using a fully convolutional neural network model

Takeaway

  • Results from this large-scale study using imaging data from a phase 3 randomized trial in patients with multiple sclerosis (MS) suggest that fully convolutional neural networks (FCNN) can automatically segment brain tissues with high accuracy.

Why this matters

  • One of the challenges in MS is that segmentation methods are often developed and tested using small datasets, raising concerns about model generalizability.

  • This study used ~1000 magnetic resonance imaging (MRI) datasets acquired as part of the phase 3 CombiRx trial to develop a FCNN model that can automatically segment MS brain tissues with high accuracy.