An End-to-end Approach to Semantic Segmentation with 3D CNN and Posterior-CRF in Medical Images
This work addresses the challenge of improving spatial coherence in medical image segmentation for applications like white matter hyperintensities analysis, representing an incremental advancement over existing CRF methods.
The paper tackled the problem of refining voxel classification in medical image segmentation by proposing an end-to-end training method called Posterior-CRF, which applies a 3D fully connected CRF to CNN posterior probabilities and optimizes both components together, resulting in outperformance over CNN, post-processing CRF, and other end-to-end CRF approaches in white matter hyperintensities segmentation.
Fully-connected Conditional Random Field (CRF) is often used as post-processing to refine voxel classification results by encouraging spatial coherence. In this paper, we propose a new end-to-end training method called Posterior-CRF. In contrast with previous approaches which use the original image intensity in the CRF, our approach applies 3D, fully connected CRF to the posterior probabilities from a CNN and optimizes both CNN and CRF together. The experiments on white matter hyperintensities segmentation demonstrate that our method outperforms CNN, post-processing CRF and different end-to-end training CRF approaches.