Abstract
AbstractUnlike existing fully-supervised approaches, we rethink colorectal polyp segmentation from an out-of-distribution perspective with a simple but effective self-supervised learning approach. We leverage the ability of masked autoencoders–self-supervised vision transformers trained on a reconstruction task–to learn in-distribution representations, here, the distribution of healthy colon images. We then perform out-of-distribution reconstruction and inference, with feature space standardisation to align the latent distribution of the diverse abnormal samples with the statistics of the healthy samples. We generate per-pixel anomaly scores for each image by calculating the difference between the input and reconstructed images and use this signal for out-of-distribution (i.e., polyp) segmentation. Experimental results on six benchmarks show that our model has excellent segmentation performance and generalises across datasets. Our code is publicly available at https://github.com/GewelsJI/Polyp-OOD.
Publisher
Springer Science and Business Media LLC
Subject
Applied Mathematics,Artificial Intelligence,Computer Networks and Communications,Computer Science Applications,Computer Vision and Pattern Recognition,Modeling and Simulation,Signal Processing,Control and Systems Engineering
Reference53 articles.
1. M. M. Center, A. Jemal, R. A. Smith, E. Ward. Worldwide variations in colorectal cancer. CA: A Cancer Journal for Clinicians, vol.59, no.6, pp.366–378, 2009. DOI: https://doi.org/10.3322/caac.20038.
2. American Cancer Society. Survival rates for colorectal cancer, [Online], Available:https://www.cancer.org/cancer/colon-rectal-cancer/detection-diagnosis-staging/survival-rates.html, 2023.
3. W. C. Zhang, C. Fu, Y. Zheng, F. Y. Zhang, Y. 1. Zhao, C. W. Sham. HSNet: A hybrid semantic network for polyp segmentation. Computers in Biology and Medicine, vol.150, Article number 106173, 2022. DOI: https://doi.org/10.1016/j.compbiomed.2022.106173.
4. D. P. Fan, G. P. Ji, T. Zhou, G. Chen, H. Z. Fu, J. B. Shen, L. Shao. PraNet: Parallel reverse attention network for polyp segmentation. In Proceedings of the 23rd International Conference on Medical Image Computing and Computer Assisted Intervention, Lima, Peru, pp. 263–273, 2020. DOI: https://doi.org/10.1007/978-3-030-59725-2_26.
5. G. P. Ji, G. B. Xiao, Y. C. Chou, D. P. Fan, K. Zhao, G. Chen, L. Van Gool. Video polyp segmentation: A deep learning perspective. Machine Intelligence Research, vol.19, no. 6, pp. 531–549, 2022. DOI: https://doi.org/10.1007/s11633-022-1371-y.