GloSAGE improves pathology model with knowledge-based text and self-supervision
Knowledge-structured and evidence-aware semantic self-distillation for glomerular pathology foundation model
Computer Vision and Pattern Recognition
Summary
Understanding detailed diseases in kidney tissue images is hard without clear labels to teach computers. The authors created GloSAGE, a method that uses existing models to label images with specific lesion details, then turns those labels into structured text descriptions. This helps the model learn better connections between images and their meanings without needing expensive manual labeling. They tested it on large kidney image datasets and showed it recognizes lesions better and helps with related diagnostic tasks.
What this means in practice
- •For pathology ai developers: Enhance kidney pathology diagnosis models by integrating semantic lesion descriptions without costly manual text annotations.
- •For medical imaging software teams: Improve image-based retrieval and classification of kidney disease features using models trained with self-generated structured pathology text.
Authors
Qiming He, Jing Li, Zihan Wang, Tian Guan, Yifei Ma, Zimo Zhao, Yanxia Wang, Hongjing Chen, Yexing Zhang, Yizhi Wang, Xinrui Chen, Xitong Ling, Minxi Ouyang, Yujie Xie, Lianghui Zhu, Yiqing Liu, Qingxia Hou, Shuyan Zhao, Xiaoqin Wang, Lili Ma, Qiang Huang, Zhiyuan Shen, Junru Cheng, Siqi Zeng, Chao He, Yonghong He, Zhe Wang
Abstract
Visual foundation models learn rich pathological morphology but lack explicit concept-level supervision. Vision--language pretraining can provide this supervision but requires costly image--text pairs. Here, we introduce GloSAGE, a semantic self-distillation framework that uses lesion profiles predicted by a pretrained visual teacher to construct structured pathology descriptions. Reliability-guided selection identifies image--text anchors, while Knowledge-Structured Pathology Text Synthesis and Evidence-Guided Negative-pair Construction account for partially observed lesions. Image--text alignment trains projection layers while retaining the frozen visual encoder; an intermediate Semantic Transfer Layer supplies features for downstream tasks. Developed on GloMillion and evaluated across seven glomerulus- and case-level cohorts, GloSAGE supported zero-shot lesion recognition, image--text retrieval, hard-negative discrimination and concept-intervention analyses. It outperformed its source visual model in most lesion-recognition tasks and improved lesion grading, while supporting competitive case-level disease classification, injury grading and baseline renal function estimation. Ablations supported the combined contributions of knowledge-structured text and evidence-guided negatives, and the intermediate transfer layer improved mean lesion-recognition ROC-AUC relative to the complete projection output. A frozen language model conditioned on GloSAGE features generated glomerular descriptions. These findings suggest that teacher-derived lesion profiles can provide annotation-efficient semantic supervision for extending a glomerular visual foundation model.