Academic radiology · 2026
Quantifying the Heterogeneity of TRUS-Visible Lesions for Predicting Prostate Cancer: A Multicenter Preliminary Study.
СтатьяМногоцентровое исследованиеНазвание и аннотация: Перевод в очереди
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.
Zheng Bing, Zhu Jia, Xu Min, Wu Xia, Jiang Jie, Zhang Yong, Shen Cheng, Zhang Zhongxin, Huang Haiming, Liu Xiang
Аннотация
RATIONALE AND OBJECTIVES: This study aimed to quantitatively characterize the heterogeneity of Transrectal ultrasound (TRUS)-visible lesions using subregional imaging analysis, and to develop and validate an intratumoral heterogeneity (ITH) model for differentiating prostate cancer (PCa) from benign lesions.
MATERIALS AND METHODS: This retrospective study included patients from three medical centers between June 2022 and April 2025 who underwent TRUS examination prior to biopsy and presented with TRUS-visible lesions. Lesions were partitioned into multiple imaging subregions using K-means clustering, and radiomic features were extracted from both the whole lesion and each subregion. Four predictive models were constructed using multiple machine-learning classifiers: a whole-lesion model (Intratumor) and three subregional models (HabitatH1, HabitatH2, and HabitatH3). An ITH model was developed by integrating the output probabilities of the subregional models. The diagnostic performance of each model was systematically evaluated. In addition, the performance of the ITH model was compared with that of radiologists with different levels of experience.
RESULTS: Among all subregional models, the model integrating HabitatH2 and HabitatH3 demonstrated superior performance and was defined as the ITH model. the ITH model exhibited an AUC of 0.879 (95% CI: 0.808-0.950) in the internal validation set and 0.818 (95%CI: 0.715-0.921) in the external validation set, respectively, outperforming the performance of both the Intratumor and clinical models. The ITH model demonstrated superior diagnostic performance compared to junior radiologists (p<0.05) and significantly enhanced the diagnostic accuracy of the junior radiologists (p<0.05).
CONCLUSIONS: This study developed an ITH model to quantify the heterogeneity of TRUS-visible lesions, demonstrating potential predictive value for PCa.
MATERIALS AND METHODS: This retrospective study included patients from three medical centers between June 2022 and April 2025 who underwent TRUS examination prior to biopsy and presented with TRUS-visible lesions. Lesions were partitioned into multiple imaging subregions using K-means clustering, and radiomic features were extracted from both the whole lesion and each subregion. Four predictive models were constructed using multiple machine-learning classifiers: a whole-lesion model (Intratumor) and three subregional models (HabitatH1, HabitatH2, and HabitatH3). An ITH model was developed by integrating the output probabilities of the subregional models. The diagnostic performance of each model was systematically evaluated. In addition, the performance of the ITH model was compared with that of radiologists with different levels of experience.
RESULTS: Among all subregional models, the model integrating HabitatH2 and HabitatH3 demonstrated superior performance and was defined as the ITH model. the ITH model exhibited an AUC of 0.879 (95% CI: 0.808-0.950) in the internal validation set and 0.818 (95%CI: 0.715-0.921) in the external validation set, respectively, outperforming the performance of both the Intratumor and clinical models. The ITH model demonstrated superior diagnostic performance compared to junior radiologists (p<0.05) and significantly enhanced the diagnostic accuracy of the junior radiologists (p<0.05).
CONCLUSIONS: This study developed an ITH model to quantify the heterogeneity of TRUS-visible lesions, demonstrating potential predictive value for PCa.