Caner Ercan

Caner Ercan

Full-stack
Computational Pathologist

Publications & Research

Below is an overview of my research programme, organised around three thematic pillars. You can also explore my full publication record on Web of Science and Google Scholar.

First-Author Projects

Barrett's Esophagus Path-Omics

Path-Omics on Barrett's Oesophagus

Weakly supervised deep learning predicts DNA content abnormalities directly from routine H&E biopsy images.

View project →
BEACON Framework

BEACON: Spatial Immune-Molecular Profiling

Integrative framework for cancer risk stratification from routine Barrett's oesophagus histopathology.

View project →
IHC-TIME

IHC-TIME: HCC Immune Microenvironment

Computational characterisation of spatial immune phenotypes with prognostic significance in hepatocellular carcinoma.

View project →
AI(H)

AI(H): Autoimmune Hepatitis Assessment

Multi-task deep learning model for quantitative, reproducible grading and staging of autoimmune hepatitis from H&E slides.

View project →
FAP in CRC

FAP Expression in Colorectal Cancer

Stromal FAP drives immune microenvironment remodelling toward an immunosuppressive state in colorectal cancer.

View project →
FNH to HCC

Malignant Potential of Focal Nodular Hyperplasia

First genomic evidence that a benign hepatic lesion can give rise to hepatocellular carcinoma through clonal selection.

View project →

Research Themes

Spatially-Aware Deep Learning for Histopathological Tissue Characterisation

Combining pathology expertise with proficiency in machine learning and computational image analysis, I developed and led projects building quantitative, automated tools for histopathological tissue assessment that access biological information beyond the reach of manual review. My most significant contribution was a weakly supervised deep learning framework that predicts DNA content abnormalities, a key molecular marker of cancer progression, directly from routine H&E whole-slide images of Barrett’s oesophagus biopsies, introducing a novel multi-instance learning augmentation strategy that substantially improved model performance. I further developed AI(H), a multi-task deep learning model for comprehensive histological assessment of autoimmune hepatitis that characterises multiple tissue compartments and pathological features from routine staining, providing spatially resolved, quantitative, and interpretable tissue profiles that improve reproducibility. Complementing these tools, I contributed to establishing an expert consensus morphological framework for standardised tumour-infiltrating lymphocyte assessment in gastro-oesophageal carcinomas, defining the evaluation criteria required for clinical validation of immune biomarkers in this cancer type.

  1. Ercan C, Pan X, Paulson TG, et al. Predicting DNA content abnormalities in Barrett’s oesophagus: a weakly supervised learning paradigm. MIDL. 2024. Project pageDOI
  2. Ercan C, Kordy K, Knuuttila A, et al. A deep-learning-based model for assessment of autoimmune hepatitis from histology: AI(H). Virchows Arch. 2024;485(6):1095-1105. Project pageDOI
  3. Weeda YA, Salgado R, van Herpe F, …, Ercan C, …, Meijer SL. Making sense of TILs: recommendations for morphological assessment of tumour-infiltrating lymphocytes in gastro-oesophageal carcinoma. Histopathology. 2026. DOI

Spatial Profiling of Immune, Stromal, and Tissue Features for Cancer Progression Risk

I established a programme of work linking computational spatial profiling to cancer progression biology across multiple cancer and precancer systems. In my primary PhD project, I pioneered the computational characterisation of the spatial immune microenvironment in hepatocellular carcinoma, developing a deep learning pipeline that classifies tumour immunophenotypes from immunohistochemistry slides and demonstrating that computationally derived immunophenotypes carry prognostic significance. In a parallel project, I co-led a study establishing that stromal activation in colorectal cancer drives immune microenvironment remodelling toward an immunosuppressive state, linking stromal biology to immune tolerance and adverse prognosis. My most advanced contribution in this area is BEACON, a spatially aware integrative framework that jointly characterises molecular abnormality and spatial immune ecology from routine Barrett’s oesophagus histopathology, enabling integrated cancer risk stratification from routine pathology images. I further contributed to identifying a morphometric tissue composition signature predicting freedom from invasive progression in ductal carcinoma in situ, demonstrating the generalisability of computational spatial profiling for precancer risk assessment.

  1. Ercan C, Renne SL, Di Tommaso L, Ng CK, Piscuoglio S, Terracciano LM. Hepatocellular carcinoma immune microenvironment analysis: a comprehensive assessment with computational and classical pathology. Clin Cancer Res. 2024;30(22):5105-5115. Project pageDOI
  2. Coto-Llerena M*, Ercan C*, Kancherla V, et al. High expression of FAP in colorectal cancer is associated with angiogenesis and immunoregulation processes. Front Oncol. 2020;10:979. Project pageDOI
  3. Ercan C, Pan X, Paulson TG, et al. Histopathology-based spatial profiling of immune and molecular features predicts cancer risk in Barrett’s oesophagus. medRxiv. 2025. Project pageDOI
  4. Sobral-Leite M, Castillo SP, Vonk S, …, Ercan C, …, Wesseling J. A morphometric signature to identify ductal carcinoma in situ with a low risk of progression. NPJ Precis Oncol. 2025;9(1):25. DOI

Clonal Evolution, Tumour Heterogeneity, and Tissue Dynamics in Carcinogenesis

By integrating expertise in histopathology, molecular pathology, and translational research, I conducted and contributed to several studies investigating the molecular dynamics of tumour heterogeneity, clonal evolution, and tissue cell fate across carcinogenesis and organ regeneration. In the most conceptually significant of these contributions, I provided the first genomic demonstration that focal nodular hyperplasia, a hepatic lesion regarded as definitively benign, can give rise to hepatocellular carcinoma through clonal selection, establishing a shared genomic ancestry between the precursor lesion and the adjacent carcinoma. I further contributed to the molecular dissection of intra-tumoral heterogeneity in pulmonary pleomorphic carcinoma through multi-region profiling of histologically distinct tumour components, establishing the evolutionary relationships between tumour components and the molecular programs driving their divergent phenotypes. Extending this to regenerative biology, I contributed to the discovery of bipotent transitional liver progenitor cells, demonstrating through dual genetic lineage tracing that a biliary cell-derived progenitor population retains the capacity to adopt either hepatocyte or biliary fate during severe liver injury.

  1. Ercan C, Coto-Llerena M, Gallon J, et al. Genomic analysis of focal nodular hyperplasia with associated hepatocellular carcinoma unveils its malignant potential: a case report. Commun Med. 2022;2(1):11. Project pageDOI
  2. Roma L, Ercan C, Conticelli F, et al. Tracing tumour heterogeneity of pleomorphic carcinoma of the lung. J Thorac Oncol. 2024;19(9):1284-1296. DOI
  3. Pu W, Zhu H, Zhang M, …, Ercan C, …, Zhou B. Bipotent transitional liver progenitor cells contribute to liver regeneration. Nat Genet. 2023;55(4):651-664. DOI

Collaborations & Selected Co-authored Work

Postdoc Period (MD Anderson / CHUV)

PhD Period (University of Basel)