Selective labeling: identifying representative sub-volumes for interactive segmentation

Luengo, Imanol; Basham, Mark; French, Andrew P.

Abstract

Automatic segmentation of challenging biomedical volumes with multiple objects is still an open research field. Automatic approaches usually require a large amount of training data to be able to model the complex and often noisy appearance and structure of biological organelles and their boundaries. However, due to the variety of different biological specimens and the large volume sizes of the datasets, training data is costly to produce, error prone and sparsely available. Here, we propose a novel Selective Labeling algorithm to overcome these challenges; an unsupervised sub-volume proposal method that identifies the most representative regions of a volume. This massively-reduced subset of regions are then manually labeled and combined with an active learning procedure to fully segment the volume. Results on a publicly available EM dataset demonstrate the quality of our approach by achieving equivalent segmentation accuracy with only 5 % of the training data.

Citation

Luengo, I., Basham, M., & French, A. P. (2016). Selective labeling: identifying representative sub-volumes for interactive segmentation. In Patch-based Techniques in Medical Imaging (17-24). https://doi.org/10.1007/978-3-319-47118-1_3

• Conference Name: Second International Workshop on Patch-based Techniques in Medical Imaging
• Conference Location: Athens, Greece
• Start Date: Oct 17, 2016
• Acceptance Date: Sep 22, 2016
• Online Publication Date: Sep 21, 2016
• Publication Date: Sep 22, 2016
• Deposit Date: Aug 11, 2017
• Publicly Available Date: Aug 11, 2017
• Journal: Lecture Notes in Computer Science
• Electronic ISSN: 0302-9743
• Peer Reviewed: Peer Reviewed
• Volume: 9993
• Pages: 17-24
• Series ISSN: 1611-3349
• Book Title: Patch-based Techniques in Medical Imaging
• ISBN: 978-3-319-47117-4
• DOI: https://doi.org/10.1007/978-3-319-47118-1_3
• Keywords: Unsupervised; Sub-volume proposals; Interactive segmentation; Active learning; Affinity clustering; Supervoxels
• Public URL: https://nottingham-repository.worktribe.com/output/810009
• Publisher URL: https://link.springer.com/chapter/10.1007/978-3-319-47118-1_3
• Additional Information: Part of: Patch-based techniques in medical imaging:; second International Workshop, Patch-MI 2016, held in conjunction with MICCAI 2016, Athens, Greece, October 17, 2016, proceedings