Context-aware virtual adversarial training for anatomically-plausible segmentation
This addresses the issue for clinicians who need reliable segmentation outputs in medical imaging, though it is incremental as it builds on existing adversarial training methods.
The paper tackled the problem of deep neural networks producing anatomically implausible segmentations, such as those with holes or disconnected regions, by proposing a Context-aware Virtual Adversarial Training method that enforces topological constraints like connectivity, resulting in accurate and anatomically-plausible segmentations on two clinically-relevant datasets.
Despite their outstanding accuracy, semi-supervised segmentation methods based on deep neural networks can still yield predictions that are considered anatomically impossible by clinicians, for instance, containing holes or disconnected regions. To solve this problem, we present a Context-aware Virtual Adversarial Training (CaVAT) method for generating anatomically plausible segmentation. Unlike approaches focusing solely on accuracy, our method also considers complex topological constraints like connectivity which cannot be easily modeled in a differentiable loss function. We use adversarial training to generate examples violating the constraints, so the network can learn to avoid making such incorrect predictions on new examples, and employ the Reinforce algorithm to handle non-differentiable segmentation constraints. The proposed method offers a generic and efficient way to add any constraint on top of any segmentation network. Experiments on two clinically-relevant datasets show our method to produce segmentations that are both accurate and anatomically-plausible in terms of region connectivity.