Domain Game: Disentangle Anatomical Feature for Single Domain Generalized Segmentation
This addresses the out-of-distribution generalization problem in medical imaging, where only one source domain is available, by improving feature disentanglement to enhance segmentation accuracy across different sites.
The paper tackles the challenge of single domain generalization for medical image segmentation by proposing a framework that disentangles diagnostic features from domain-specific ones using geometric transformations, achieving an 11.8% performance boost in prostate segmentation and 10.5% in brain tumor segmentation compared to the second-best method.
Single domain generalization aims to address the challenge of out-of-distribution generalization problem with only one source domain available. Feature distanglement is a classic solution to this purpose, where the extracted task-related feature is presumed to be resilient to domain shift. However, the absence of references from other domains in a single-domain scenario poses significant uncertainty in feature disentanglement (ill-posedness). In this paper, we propose a new framework, named \textit{Domain Game}, to perform better feature distangling for medical image segmentation, based on the observation that diagnostic relevant features are more sensitive to geometric transformations, whilist domain-specific features probably will remain invariant to such operations. In domain game, a set of randomly transformed images derived from a singular source image is strategically encoded into two separate feature sets to represent diagnostic features and domain-specific features, respectively, and we apply forces to pull or repel them in the feature space, accordingly. Results from cross-site test domain evaluation showcase approximately an ~11.8% performance boost in prostate segmentation and around ~10.5% in brain tumor segmentation compared to the second-best method.