PrPSeg: Universal Proposition Learning for Panoramic Renal Pathology Segmentation

• Published in: Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) Vol. 2024; pp. 11736 - 11746
• Main Authors: Deng, Ruining, Liu, Quan, Cui, Can, Yao, Tianyuan, Yue, Jialin, Xiong, Juming, Yu, Lining, Wu, Yifei, Yin, Mengmeng, Wang, Yu, Zhao, Shilin, Tang, Yucheng, Yang, Haichun, Huo, Yuankai
• Format: Conference Proceeding Journal Article
• Language: English
• Published: United States IEEE 01.06.2024
• Subjects: Computational modeling; digital pathology; dynamic network; Head; hierarchical taxonomy; Image segmentation; Pathology; segmentation; semi-supervised learning; Semisupervised learning; Training; Transforms
• ISSN: 1063-6919; 1063-6919; 2575-7075
• DOI: 10.1109/CVPR52733.2024.01115

Understanding the anatomy of renal pathology is crucial for advancing disease diagnostics, treatment evaluation, and clinical research. The complex kidney system comprises various components across multiple levels, including regions (cortex, medulla), functional units (glomeruli, tubules), and cells (podocytes, mesangial cells in glomeru-lus). Prior studies have predominantly overlooked the in-tricate spatial interrelations among objects from clinical knowledge. In this research, we introduce a novel univer-sal proposition learning approach, called panoramic re-nal pathology segmentation (PrPSeg), designed to segment comprehensively panoramic structures within kidney by in-tegrating extensive knowledge of kidney anatomy. In this paper, we propose (1) the design of a comprehensive universal proposition matrix for renal pathol-ogy, facilitating the incorporation of classification and spatial relationships into the segmentation process; (2) a token-based dynamic head single network architecture, with the improvement of the partial label image segmentation and capability for future data enlargement; and (3) an anatomy loss function, quantifying the inter-object relation-ships across the kidney.