Semi-supervised Learning for Biomedical Image Segmentation via Forest Oriented Super Pixels(Voxels)
In this paper, we focus on semi-supervised learning for biomedical image segmentation, so as to take advantage of huge unlabelled data. We observe that there usually exist some homogeneous connected areas of low confidence in biomedical images, which tend to confuse the classifier trained with limited labelled samples. To cope with this difficulty, we propose to construct forest oriented super pixels(voxels) to augment the standard random forest classifier, in which super pixels(voxels) are built upon the forest based code. Compared to the state-of-the-art, our proposed method shows superior segmentation performance on challenging 2D/3D biomedical images. The full implementation (based on Matlab) is available at https://github.com/lingucv/ssl_superpixels.
KeywordsImage segmentation Semi-supervised learning Random forest Super pixels(voxels)
This work was funded by ImPACT Program of Council for Science, Technology and Innovation (Cabinet Office, Government of Japan).
- 4.Coye, T.: A novel retinal blood vessel segmentation algorithm for fundus images. MATLAB Central File Exchange, January 2017Google Scholar
- 5.Gu, L., Cheng, L.: Learning to boost filamentary structure segmentation. In: ICCV 2015 (2015)Google Scholar
- 6.Jaakkola, M.S.T., Szummer, M.: Partially labeled classification with Markov random walks. In: NIPS 2002, vol. 14, pp. 945–952 (2002)Google Scholar
- 8.Liu, Y.J., Yu, C.C., Yu, M.J., He, Y.: Manifold SLIC: a fast method to compute content-sensitive superpixels. In: CVPR 2016, June 2016Google Scholar
- 9.Ren, S., Cao, X., Wei, Y., Sun, J.: Global refinement of random forest. In: CVPR 2015 (2015)Google Scholar
- 13.Yang, Z., Cohen, W.W., Salakhutdinov, R.: Revisiting semi-supervised learning with graph embeddings. In: ICML (2016)Google Scholar