Zero-shot benchmark code: histopathology foundation models in canine oncology (CONCH, Pathos-Gemma4) - MIDOG++
Year: 2026
Type: software
Abstract: Evaluation code for the zero-shot benchmark of histopathology foundation models (CONCH and Pathos-Gemma4) on the MIDOG++ canine subset (148 slides; mast cell tumour, lymphoma, pulmonary carcinoma). Includes CONCH patch-embedding benchmark, Pathos-Gemma4 llama.cpp inference scripts, and head-to-head comparison scripts with result files. Accompanies the manuscript submitted to Medical Image Analysis.
Authors: Villanueva-Castillo, Arnulfo, Moreno-Mejia, Ruby Sandy, Cruz-Espinosa, Alexis, Mancilla-Simbro, Claudia, Castro-Bautista, Briseida L., Fernandez-Meneses, Erick C., Zambrano-Gonzalez, Miguel A., Rodriguez-Andrade, Fabiola, Cruz-Avina, Juan R., Hinojosa-Moya, J. Jesus, Lucio-Castillo, Hermilo, Pastelin-Rojas, Cesar F., Ramirez-Mata, Alberto
Topic: AI in cancer detection
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7220992012