Cancer cytopathology · 2026
Artificial intelligence in nongynecologic cytology: A systematic review of current research and commercial tools.
СтатьяСистематический обзорПоддержка исследования: вне правительства СШАНазвание 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.
Byambadorj Tsandmaa, Alam Mohammad Rizwan, Chong Yosep
Аннотация
BACKGROUND: Nongynecologic (non-GYN) cytology plays an important role in cellular and molecular cancer diagnosis, although its use is limited by variability and interobserver discrepancies. Recent studies suggest that artificial intelligence (AI) may improve diagnostic consistency and performance.
BACKGROUND: This systematic review evaluated contemporary AI research in non-GYN cytology and examined commercially available systems.
METHODS: MEDLINE, Embase, and the Cochrane Library were searched for English-language studies published from January 2010 to September 2024. Of ∼24,000 records screened, 71 met inclusion criteria. Commercial platforms (KFBIO, Landing Med, VitaDx/VisioCyt, AIxMED, and CellsVision) were assessed using publicly available regulatory documents, technical specifications, and vendor-reported validation data.
RESULTS: Many studies reported high internal diagnostic performance, often exceeding 90% accuracy. Patch-level performance often reached ∼94%, exceeding whole slide image-level metrics. Reduced interobserver variability was frequently observed, with accuracies of 95.9% for senior and 94.4% for junior pathologists. Several studies documented shorter diagnostic time. Thyroid (37%) and urinary bladder (30%) cytology were most frequently studied. AI applications included categorization, segmentation, and atypia detection, with recent work exploring immunocytochemistry support, mutation-associated pattern recognition, and indirect prediction of histologic features.
CONCLUSION: AI in non-GYN cytology shows promise but remains limited by data set quality, weak external validation, and regulatory barriers. At present, AI tools function primarily as decision-support systems. Advancing clinical adoption will require multicenter validation, standardized data sets, and careful integration with expert interpretation.
BACKGROUND: This systematic review evaluated contemporary AI research in non-GYN cytology and examined commercially available systems.
METHODS: MEDLINE, Embase, and the Cochrane Library were searched for English-language studies published from January 2010 to September 2024. Of ∼24,000 records screened, 71 met inclusion criteria. Commercial platforms (KFBIO, Landing Med, VitaDx/VisioCyt, AIxMED, and CellsVision) were assessed using publicly available regulatory documents, technical specifications, and vendor-reported validation data.
RESULTS: Many studies reported high internal diagnostic performance, often exceeding 90% accuracy. Patch-level performance often reached ∼94%, exceeding whole slide image-level metrics. Reduced interobserver variability was frequently observed, with accuracies of 95.9% for senior and 94.4% for junior pathologists. Several studies documented shorter diagnostic time. Thyroid (37%) and urinary bladder (30%) cytology were most frequently studied. AI applications included categorization, segmentation, and atypia detection, with recent work exploring immunocytochemistry support, mutation-associated pattern recognition, and indirect prediction of histologic features.
CONCLUSION: AI in non-GYN cytology shows promise but remains limited by data set quality, weak external validation, and regulatory barriers. At present, AI tools function primarily as decision-support systems. Advancing clinical adoption will require multicenter validation, standardized data sets, and careful integration with expert interpretation.