Automatic segmentation of spinal multiple sclerosis lesions: How to generalize across MRI contrasts?
This work addresses segmentation challenges for spinal MS lesions in medical imaging, but it is incremental as it reveals a bottleneck rather than achieving generalization across contrasts.
The study tackled the problem of generalizing spinal multiple sclerosis lesion segmentation across MRI contrasts using Feature-wise Linear Modulation (FiLM), but found that a well-optimized U-Net matched its performance with a Dice score of 0.72, indicating a bottleneck likely due to inter-rater variability estimated at 0.61 Dice score.
Despite recent improvements in medical image segmentation, the ability to generalize across imaging contrasts remains an open issue. To tackle this challenge, we implement Feature-wise Linear Modulation (FiLM) to leverage physics knowledge within the segmentation model and learn the characteristics of each contrast. Interestingly, a well-optimised U-Net reached the same performance as our FiLMed-Unet on a multi-contrast dataset (0.72 of Dice score), which suggests that there is a bottleneck in spinal MS lesion segmentation different from the generalization across varying contrasts. This bottleneck likely stems from inter-rater variability, which is estimated at 0.61 of Dice score in our dataset.