[ISMB/ECCB 2025] Predicting fine-grained cell types from histology images through cross-modal learning in spatial transcriptomics
Chaoyang Yan, Zhihan Ruan, Songkang Chen, Yichen Pan, Xue Han, Yuanyu Li, Jian Liu
[Paper: TBD][Code]
A cross-modal unified representation learning framework (CUCA) is proposed for identifying fine-grained cell types from histology images. CUCA is trained on paired morphology-molecule spatial transcriptomics data, enabling it to infer fine-grained cell abundances solely from pathology images.

[ISMB 2024] PhiHER2: Phenotype-informed weakly supervised model for HER2 status prediction from pathological images
Chaoyang Yan, Jialiang Sun, Yiming Guan, Jiuxin Feng, Hong Liu, Jian Liu
[Paper][Code]
PhiHER2 is a phenotype-informed multiple instance learning architecture for the prediction of the HER2 molecular status from pathological images of breast cancer. It provides valuable insights into the heterogeneity of morphological patterns associated with molecular biomarkers.

[Fundamental Research 2022] Deep learning framework for comprehensive molecular and prognostic stratifications of triple-negative breast cancer
Shen Zhao*, Chaoyang Yan*, Hong Lv*, et.al
[Paper]
A deep learning-based framework is devised for comprehensive predictions of molecular features, subtypes and prognosis from pathological WSIs. It enables comprehensive stratifications of TNBC patients and provide valuable information for therapeutic decision-making.

[Magnetic Resonance in Medicine 2022] Scale- and Slice-aware Net (S2aNet) for 3D segmentation of organs and musculoskeletal structures in pelvic MRI
Chaoyang Yan, Jing-Jing Lu, Kang Chen, Lei Wang, Haoda Lu, Li Yu, Mengyan Sun, Jun Xu
[Paper][Code]
S2aNet is the first model to achieve 3D dense segmentation for 54 organs and musculoskeletal structures in female pelvic MR images. It is an efficient, accurate, and automated segmentation model that can be extended to the 3D digital anatomy for personalized body.

[ISBI 2020] Prior-aware CNN with multi-task learning for colon images analysis
Chaoyang Yan, Jun Xu, Jiawei Xie, Chengfei Cai, Haoda Lu
[Paper][Code]
A prior-aware CNN framework with multi-task learning is proposed for automatic gland segmentation and images grading diagnosis simultaneously. It makes pathological image analysis more applicable and interpretable inspired by the clinical pathological diagnostic criteria.

[Computer Methods and Programs in Biomedicine 2020] Automated gleason grading on prostate biopsy slides by statistical representations of homology profile
Chaoyang Yan, Kazuaki Nakane, Xiangxue Wang, et.al
[Paper]
SRHP is a novel Statistical Representation of Homology Profile approach for automated Gleason grading on prostate biopsy slides. It captures the topological characteristics and the degree of connectivity between nuclei around the gland. These characteristics of homology profile are interpretable, visually meaningful and highly consistent with the rubric employed by pathologists.