利用一个用于病理学研究的强大开源工具实现肝活检分割自动化：HOTSPoT模型。
Automating liver biopsy segmentation with a robust, open-source tool for pathology research: the HOTSPoT model.
作者信息
Cazzaniga Giorgio, L'Imperio Vincenzo, Bonoldi Emanuela, Londoño María-Carlota, Madaleno João, Cipriano Augusta, Gevers Tom J G, Samarska Iryna V, Koc Özgür M, Villamil Alejandra, Sanchez Maria Florencia, Calvaruso Vincenza, Quattrocchi Alberto, Cabibi Daniela, Engel Bastian, Malinverno Federica, Merelli Elisa, Cristoferi Laura, Carbone Marco, Pagni Fabio, Invernizzi Pietro, Gerussi Alessio
机构信息
Department of Medicine and Surgery, Pathology, Fondazione IRCCS San Gerardo dei Tintori, University of Milano-Bicocca, Monza, Italy.
U.O.C. Anatomia Patologica e Citogenetica, Department of Hematology, Oncology and Molecular Medicine, ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy.
出版信息
NPJ Digit Med. 2025 Jul 18;8(1):455. doi: 10.1038/s41746-025-01870-1.
Artificial intelligence applications in liver pathology remain limited, with existing tools either narrowly focused or lacking external validation. This study introduces HOTSPoT, an open-source, validated transformer-based model for automated segmentation of portal tracts in H&E-stained liver biopsy whole slide images. A multi-institutional dataset of 223 cases was used, with annotations by expert hepatopathologists. HOTSPoT achieved high performance with mean Dice scores of 0.92 (train/val) and 0.91 (test), and mean IoUs of 0.86, 0.85, and 0.84, respectively, showing minimal domain shift. Automated portal tract quantification showed strong concordance with manual assessments (κ up to 0.90), and portal area correlated with fibrosis stage (r = 0.87, p < 0.001). The model is available as a TorchScript file with a modified WSInfer library, enabling efficient WSI-level inference and integration with QuPath for advanced pathology analysis.
人工智能在肝脏病理学中的应用仍然有限，现有工具要么关注范围狭窄，要么缺乏外部验证。本研究介绍了HOTSPoT，这是一种基于开源、经过验证的变压器模型，用于在苏木精和伊红（H&E）染色的肝脏活检全切片图像中自动分割门静脉。使用了一个包含223例病例的多机构数据集，并由专业肝脏病理学家进行注释。HOTSPoT表现出色，训练/验证集的平均Dice分数为0.92，测试集为0.91，平均交并比分别为0.86、0.85和0.84，显示出最小的域偏移。自动门静脉定量与手动评估高度一致（κ高达0.90），门静脉面积与纤维化阶段相关（r = 0.87，p < 0.001）。该模型以带有修改后的WSInfer库的TorchScript文件形式提供，可实现高效的全切片图像级推理，并与QuPath集成以进行高级病理学分析。
Zotero 插件