Gland surgery · 2026
Artificial intelligence for prostate cancer detection and risk stratification using transrectal ultrasound: a narrative review.
СтатьяОбзорНазвание 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.
Zhang Qiong, Zhou Chenyun, Chen Yang, Luo Yan
Аннотация
BACKGROUND AND OBJECTIVE: Prostate cancer (PCa) management and outcomes are dependent on risk stratification. Indolent disease is often managed with active surveillance, whereas clinically significant PCa (CSPCa) necessitates prompt intervention due to its aggressive potential. Although transrectal ultrasound (TRUS) is central to diagnosis and biopsy guidance, its limited resolution and high interobserver variability complicate accurate risk assessment. Artificial intelligence (AI) offers a promising solution to these limitations. This review evaluates the current landscape of TRUS-based AI models for three critical objectives: PCa detection, CSPCa identification, and risk stratification.
METHODS: We systematically searched the PubMed and Web of Science databases for peer-reviewed, original English-language articles published from 1999 to 2026.
KEY CONTENT AND FINDINGS: TRUS-based AI models have advanced significantly, achieving area under the curve (AUC) values of 0.78-0.96 for PCa detection and 0.85-0.90 for CSPCa identification, particularly when leveraging three-dimensional (3D) architectures or multiparametric fusion (e.g., elastography or contrast enhancement). Performance is robust for binary risk stratification (e.g., low-intermediate vs. high-risk). However, a critical gap remains: no existing AI model has successfully predicted the full spectrum of the five-tier International Society of Urological Pathology (ISUP) Grade Group (GG) stratification using TRUS imaging alone. Key barriers to clinical translation include challenges in precise lesion localization, the complexity of annotating risk-stratified labels, and the predominance of single-center retrospective datasets.
CONCLUSIONS: TRUS-based AI demonstrates high accuracy for PCa detection and CSPCa identification, particularly with 3D architectures or multiparametric fusion. However, the inability to predict the full five-tier ISUP GG stratification represents a major unmet need. Future research should prioritize standardized multicenter data collection and advanced algorithms to address localization challenges and enable precise risk stratification, thereby facilitating clinical translation.
METHODS: We systematically searched the PubMed and Web of Science databases for peer-reviewed, original English-language articles published from 1999 to 2026.
KEY CONTENT AND FINDINGS: TRUS-based AI models have advanced significantly, achieving area under the curve (AUC) values of 0.78-0.96 for PCa detection and 0.85-0.90 for CSPCa identification, particularly when leveraging three-dimensional (3D) architectures or multiparametric fusion (e.g., elastography or contrast enhancement). Performance is robust for binary risk stratification (e.g., low-intermediate vs. high-risk). However, a critical gap remains: no existing AI model has successfully predicted the full spectrum of the five-tier International Society of Urological Pathology (ISUP) Grade Group (GG) stratification using TRUS imaging alone. Key barriers to clinical translation include challenges in precise lesion localization, the complexity of annotating risk-stratified labels, and the predominance of single-center retrospective datasets.
CONCLUSIONS: TRUS-based AI demonstrates high accuracy for PCa detection and CSPCa identification, particularly with 3D architectures or multiparametric fusion. However, the inability to predict the full five-tier ISUP GG stratification represents a major unmet need. Future research should prioritize standardized multicenter data collection and advanced algorithms to address localization challenges and enable precise risk stratification, thereby facilitating clinical translation.