Abstract:

Medical image segmentation is a fundamental task for medical image analysis and surgical planning. In recent years, UNet-based networks have prevailed in the field of medical image segmentation. However, convolutional neural networks (CNNs) suffer from limited receptive fields, which fail to model the long-range dependency of organs or tumors. Besides, these models are heavily dependent on the training of the final segmentation head. And existing methods can not well address aforementioned limitations simultaneously. Hence, in our work, we proposed a novel shape prior module (SPM), which can explicitly introduce shape priors to promote the segmentation performance of UNet-based models. The explicit shape priors consist of global and local shape priors. The former with coarse shape representations provides networks with capabilities to model global contexts. The latter with finer shape information serves as additional guidance to relieve the heavy dependence on the learnable prototype in the segmentation head. To evaluate the effectiveness of SPM, we conduct experiments on three challenging public datasets. And our proposed model achieves state-of-the-art performance. Furthermore, SPM can serve as a plug-andplay structure into classic CNNs and Transformer-based backbones, facilitating the segmentation task on different datasets.

Fig. 1 The comparison between different segmentation paradigms with explicit shape priors. a. Atlas-based models employ m ground truths from source datasets as shape priors, and construct the transformation matrix R between source and target images. Then R is applied to shape priors for achieving segmentation masks. b. Gaussian Mixture Model (GMM) gathers N (the number of segmentation classes) learnable Gaussian distributions as shape priors to deal with the per-pixel classification. c. Our model adopts N-channel learnable shape priors as additional inputs to boost the segmentation performance of deep neural networks.

Fig. 2 Qualitative comparisons of baseline and SPM on multiple tasks.

DOI: 10.1109/TMI.2024.3469214