{"ID":2844891,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.05150","arxiv_id":"2511.05150","title":"From Linear Probing to Joint-Weighted Token Hierarchy: A Foundation Model Bridging Global and Cellular Representations in Biomarker Detection","abstract":"AI-based biomarkers can infer molecular features directly from hematoxylin \u0026 eosin (H\u0026E) 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.","short_abstract":"AI-based biomarkers can infer molecular features directly from hematoxylin \u0026 eosin (H\u0026E) 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-supervis...","url_abs":"https://arxiv.org/abs/2511.05150","url_pdf":"https://arxiv.org/pdf/2511.05150v1","authors":"[\"Jingsong Liu\",\"Han Li\",\"Nassir Navab\",\"Peter J. Schüffler\"]","published":"2025-11-07T11:05:36Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\"]","methods":"[]","has_code":false}