A unified representation network for segmentation with missing modalities
This addresses a practical issue in medical imaging for clinicians and researchers, but it is incremental as it builds on existing neural network methods.
The paper tackles the problem of medical image segmentation when input modalities are missing, violating the assumption of matching train-test distributions, and finds that a unified representation network outperforms modality dropout, improving robustness to missing data.
Over the last few years machine learning has demonstrated groundbreaking results in many areas of medical image analysis, including segmentation. A key assumption, however, is that the train- and test distributions match. We study a realistic scenario where this assumption is clearly violated, namely segmentation with missing input modalities. We describe two neural network approaches that can handle a variable number of input modalities. The first is modality dropout: a simple but surprisingly effective modification of the training. The second is the unified representation network: a network architecture that maps a variable number of input modalities into a unified representation that can be used for downstream tasks such as segmentation. We demonstrate that modality dropout makes a standard segmentation network reasonably robust to missing modalities, but that the same network works even better if trained on the unified representation.