Alternative Baselines for Low-Shot 3D Medical Image Segmentation---An Atlas Perspective

Authors

  • Shuxin Wang Department of Computer Science, Xiamen University, Xiamen, China Tencent Jarvis Lab, Shenzhen, China
  • Shilei Cao Tencent Jarvis Lab, Shenzhen, China
  • Dong Wei Tencent Jarvis Lab, Shenzhen, China
  • Cong Xie Department of Computer Science, Xiamen University, Xiamen, China Tencent Jarvis Lab, Shenzhen, China
  • Kai Ma Tencent Jarvis Lab, Shenzhen, China
  • Liansheng Wang Department of Computer Science, Xiamen University, Xiamen, China Department of Digestive Diseases, School of Medicine, Xiamen University, Xiamen, China
  • Deyu Meng School of Mathematics and Statistics, Xi’an Jiaotong University, Xi’an, China
  • Yefeng Zheng Tencent Jarvis Lab, Shenzhen, China

Keywords:

Bioinformatics, Segmentation

Abstract

Low-shot (one/few-shot) segmentation has attracted increasing attention as it works well with limited annotation. State-of-the-art low-shot segmentation methods on natural images usually focus on implicit representation learning for each novel class, such as learning prototypes, deriving guidance features via masked average pooling, and segmenting using cosine similarity in feature space. We argue that low-shot segmentation on medical images should step further to explicitly learn dense correspondences between images to utilize the anatomical similarity. The core ideas are inspired by the classical practice of multi-atlas segmentation, where the indispensable parts of atlas-based segmentation, i.e., registration, label propagation, and label fusion are unified into a single framework in our work. Specifically, we propose two alternative baselines, i.e., the Siamese-Baseline and Individual-Difference-Aware Baseline, where the former is targeted at anatomically stable structures (such as brain tissues), and the latter possesses a strong generalization ability to organs suffering large morphological variations (such as abdominal organs). In summary, this work sets up a benchmark for low-shot 3D medical image segmentation and sheds light on further understanding of atlas-based few-shot segmentation.

Downloads

Published

2021-05-18

How to Cite

Wang, S., Cao, S., Wei, D., Xie, C., Ma, K., Wang, L., Meng, D., & Zheng, Y. (2021). Alternative Baselines for Low-Shot 3D Medical Image Segmentation---An Atlas Perspective. Proceedings of the AAAI Conference on Artificial Intelligence, 35(1), 634-642. Retrieved from https://ojs.aaai.org/index.php/AAAI/article/view/16143

Issue

Section

AAAI Technical Track on Application Domains