Journal Article

Generating dermatopathology reports from gigapixel whole slide images with HistoGPT

Murphree, Dennis H., Mooyaart, Antien L., Lupperger, Valerio, Litjens, Geert, Biedermann, Tilo, Ciompi, Francesco, Koch, Valentin, Wagner, Sophia J., Guo, Ruifeng Ray, Comfere, Nneka I., Kaczmarczyk, Robert, Amir, Avital L., Lefkes, Judith, Marr, Carsten, Peng, Tingying, Tran, Manuel, Schmidle, Paul, Novotny, Brenna, Braun, Stephan A., Hardway, Heather D., Geijs, Daan J., Feuchtinger, Annette, D’Amato, Marina, Wang, Chen, Eyerich, Kilian, Böhner, Alexander

2025

Overview

Histopathology is the reference standard for diagnosing the presence and nature of many diseases, including cancer. However, analyzing tissue samples under a microscope and summarizing the findings in a comprehensive pathology report is time-consuming, labor-intensive, and non-standardized. To address this problem, we present HistoGPT, a vision language model that generates pathology reports from a patient’s multiple full-resolution histology images. It is trained on 15,129 whole slide images from 6705 dermatology patients with corresponding pathology reports. The generated reports match the quality of human-written reports for common and homogeneous malignancies, as confirmed by natural language processing metrics and domain expert analysis. We evaluate HistoGPT in an international, multi-center clinical study and show that it can accurately predict tumor subtypes, tumor thickness, and tumor margins in a zero-shot fashion. Our model demonstrates the potential of artificial intelligence to assist pathologists in evaluating, reporting, and understanding routine dermatopathology cases. Machine learning models represent an opportunity for the automatic generation of histopathology reports. Here, the authors develop HistoGPT, a vision language model that can generate reports from multiple gigapixel-sized whole slide images and also predict tumour thickness, subtypes, and margins, among other diseases.

• Publisher: Nature Publishing Group UK,Nature Publishing Group,Nature Portfolio
• Subject: 14/63 / 631/114/1305 / 631/114/1564 / 631/67/2321 / 692/699/67/1813 / Artificial Intelligence / Dermatology / Dermatology - methods / Histology / Histopathology / Humanities and Social Sciences / Humans / Image Processing, Computer-Assisted - methods / Language / Machine learning / Malignancy / Medical imaging / multidisciplinary / Natural Language Processing / Pathology / Science / Science (multidisciplinary) / Skin Neoplasms - diagnosis / Skin Neoplasms - pathology / Thickness / Tumors