Tree-based Semantic Losses: Application to Sparsely-supervised Large Multi-class Hyperspectral Segmentation
This addresses the challenge of distinguishing many subtly varying tissue classes in surgical hyperspectral imaging with sparse annotations, representing a domain-specific incremental improvement.
The paper tackles the problem of sparsely-supervised large multi-class hyperspectral segmentation by introducing tree-based semantic loss functions that exploit hierarchical label organization, achieving state-of-the-art performance on a 107-class dataset and enabling effective out-of-distribution pixel detection without compromising in-distribution segmentation.
Hyperspectral imaging (HSI) shows great promise for surgical applications, offering detailed insights into biological tissue differences beyond what the naked eye can perceive. Refined labelling efforts are underway to train vision systems to distinguish large numbers of subtly varying classes. However, commonly used learning methods for biomedical segmentation tasks penalise all errors equivalently and thus fail to exploit any inter-class semantics in the label space. In this work, we introduce two tree-based semantic loss functions which take advantage of a hierarchical organisation of the labels. We further incorporate our losses in a recently proposed approach for training with sparse, background-free annotations. Extensive experiments demonstrate that our proposed method reaches state-of-the-art performance on a sparsely annotated HSI dataset comprising $107$ classes organised in a clinically-defined semantic tree structure. Furthermore, our method enables effective detection of out-of-distribution (OOD) pixels without compromising segmentation performance on in-distribution (ID) pixels.