Multi-modal segmentation with missing MR sequences using pre-trained fusion networks

TL;DR

This paper introduces a multi-modal segmentation network designed to handle missing MRI sequences by combining dropout training, multi-pathway architecture, and pre-training, improving performance on incomplete data in medical imaging.

Contribution

It presents a novel neural network architecture with specific training and fusion strategies to effectively segment images with missing modalities, advancing the handling of incomplete multi-modal data.

Findings

Significant performance improvement on missing data scenarios

Reduced memory requirements during training

Outperforms state-of-the-art methods on BraTS dataset

Abstract

Missing data is a common problem in machine learning and in retrospective imaging research it is often encountered in the form of missing imaging modalities. We propose to take into account missing modalities in the design and training of neural networks, to ensure that they are capable of providing the best possible prediction even when multiple images are not available. The proposed network combines three modifications to the standard 3D UNet architecture: a training scheme with dropout of modalities, a multi-pathway architecture with fusion layer in the final stage, and the separate pre-training of these pathways. These modifications are evaluated incrementally in terms of performance on full and missing data, using the BraTS multi-modal segmentation challenge. The final model shows significant improvement with respect to the state of the art on missing data and requires less memory during training.