Cancer medicine · 2026
Region-Based Segmentation of Lymph Node Metastases in Whole-Slide Images of Colorectal Cancer: A Pilot Clinical Study.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Fayzullin Alexey, Savelov Nikita, Balkivskiy Artur, Ivanova Elena, Timakova Anna, Funtikov Vladimir, Shelomentseva Ekaterina, Panchenko Nikita, Spiridonov Igor, Korkonishko Ivan, Makhina Mariia, Kutuzova Evgeniia, Meretukov Daur, Kretova Natalia, Demura Tatiana, Timashev Peter, Parchiev Ruslan
Аннотация
BACKGROUND: Digital technologies and artificial intelligence (AI) are transforming medical diagnostics, particularly in pathology. This study presented a two-stage computer vision model designed to detect colorectal cancer metastases in whole slide images (WSIs) of lymph nodes.
METHODS: We developed a classification-segmentation pipeline optimized for both accuracy and efficiency. The model was trained on 108 WSIs and evaluated on 554 WSIs collected from two institutions using Leica Aperio AT2 and Hamamatsu NanoZoomer S360 scanners.
RESULTS: The classification model achieved a recall of 1.0 and a specificity of 0.935, while the segmentation model reported a Dice coefficient of 0.818 ± 0.105. Pathologists appreciated the model's precision in distinguishing solitary cancer cells from histiocytosis, reducing the need for peer consultations. Feedback from the pilot study indicated that the AI tool served as a valuable second opinion, enhancing diagnostic confidence.
CONCLUSION: This study explored the practical applications of AI in clinical pathology, offering perspectives from both pathologists and data scientists. Our findings highlighted how AI can streamline workflows, improve diagnostic accuracy, and support personalized treatment planning. The integration of AI into pathology workflows has the potential to redefine diagnostic standards while maintaining the critical role of pathologists in decision-making.
METHODS: We developed a classification-segmentation pipeline optimized for both accuracy and efficiency. The model was trained on 108 WSIs and evaluated on 554 WSIs collected from two institutions using Leica Aperio AT2 and Hamamatsu NanoZoomer S360 scanners.
RESULTS: The classification model achieved a recall of 1.0 and a specificity of 0.935, while the segmentation model reported a Dice coefficient of 0.818 ± 0.105. Pathologists appreciated the model's precision in distinguishing solitary cancer cells from histiocytosis, reducing the need for peer consultations. Feedback from the pilot study indicated that the AI tool served as a valuable second opinion, enhancing diagnostic confidence.
CONCLUSION: This study explored the practical applications of AI in clinical pathology, offering perspectives from both pathologists and data scientists. Our findings highlighted how AI can streamline workflows, improve diagnostic accuracy, and support personalized treatment planning. The integration of AI into pathology workflows has the potential to redefine diagnostic standards while maintaining the critical role of pathologists in decision-making.