From Linear Probing to Joint-Weighted Token Hierarchy: A Foundation Model Bridging Global and Cellular Representations in Biomarker Detection

MCML Authors

Jingsong Liu

→ Group Peter Schüffler
 Computational Pathology

Han Li

Dr.

→ Group Nassir Navab
 Computer Aided Medical Procedures & Augmented Reality

Nassir Navab

Prof. Dr.

Principal Investigator

Computer Aided Medical Procedures & Augmented Reality

Peter Schüffler

Prof. Dr.

Associate

Computational Pathology

Abstract

AI-based biomarkers can infer molecular features directly from hematoxylin & eosin (H&E) slides, yet most pathology foundation models (PFMs) rely on global patch-level embeddings and overlook cell-level morphology. We present a PFM model, JWTH (Joint-Weighted Token Hierarchy), which integrates large-scale self-supervised pretraining with cell-centric post-tuning and attention pooling to fuse local and global tokens. Across four tasks involving four biomarkers and eight cohorts, JWTH achieves up to 8.3% higher balanced accuracy and 1.2% average improvement over prior PFMs, advancing interpretable and robust AI-based biomarker detection in digital pathology.

misc LLN+25

---

Preprint

Nov. 2025

---

Authors

J. Liu • H. Li • N. Navab • P. J. Schüffler

---

Links

arXiv

---

Research Area

 C1 | Medicine