PloS one · 2026
MR-based radiomics of mesorectal fat for improved prediction of perirectal lymph node metastasis and extramural venous invasion in locally advanced rectal cancer.
СтатьяНазвание 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.
Swaengdee Yaniga, Khongwirotphan Sararas, Lasode Jaravee, Kuecharoen Phakakarn, Phetvilay Phathayphout, Limvorapitak Thitithep, Sanpavat Anapat, Sriswasdi Sira, Boonsirikamchai Piyaporn, Rakvongthai Yothin
Аннотация
OBJECTIVE: Accurately assessing residual disease after neoadjuvant chemoradiotherapy (nCRT) in locally advanced rectal cancer (LARC) remains challenging. Residual extramural venous invasion (EMVI) and perirectal lymph node (PLN) metastasis indicate adverse outcomes, but treatment-related changes obscure their detection on post-treatment MRI. This study developed MRI-based radiomics models to predict residual EMVI and PLN metastasis using post-nCRT restaging MRI.
MATERIALS AND METHODS: In this retrospective study, 219 patients with LARC who completed nCRT and underwent post-treatment MRI for restaging prior to surgery were included. Radiomic features were extracted from manually segmented regions of interest encompassing the primary tumor and mesorectal fat on high-resolution T2-weighted images using PyRadiomics. Logistic regression (LR), support vector machine (SVM), and random forest (RF) models were developed to predict pathological EMVI and PLN status. Model performance was assessed using repeated 5-fold cross-validation, with the area under the receiver operating characteristic curve (AUC) as the primary evaluation metric. Differences in model performance were compared using DeLong test.
RESULTS: For EMVI prediction, the combined tumor and mesorectal fat radiomics model achieved the highest AUC of 0.797 ± 0.073 using the LR model. For PLN prediction, the combined model also demonstrated superior performance, achieving an AUC of 0.824 ± 0.073. Models incorporating both tumor and mesorectal fat features consistently outperformed single-region models.
CONCLUSION: MRI-based radiomics models using post-nCRT restaging images could predict residual EMVI and PLN metastasis in LARC. Incorporating mesorectal fat features improved model performance, suggesting that information from the surrounding mesorectal compartment may be useful for post-treatment risk assessment.
MATERIALS AND METHODS: In this retrospective study, 219 patients with LARC who completed nCRT and underwent post-treatment MRI for restaging prior to surgery were included. Radiomic features were extracted from manually segmented regions of interest encompassing the primary tumor and mesorectal fat on high-resolution T2-weighted images using PyRadiomics. Logistic regression (LR), support vector machine (SVM), and random forest (RF) models were developed to predict pathological EMVI and PLN status. Model performance was assessed using repeated 5-fold cross-validation, with the area under the receiver operating characteristic curve (AUC) as the primary evaluation metric. Differences in model performance were compared using DeLong test.
RESULTS: For EMVI prediction, the combined tumor and mesorectal fat radiomics model achieved the highest AUC of 0.797 ± 0.073 using the LR model. For PLN prediction, the combined model also demonstrated superior performance, achieving an AUC of 0.824 ± 0.073. Models incorporating both tumor and mesorectal fat features consistently outperformed single-region models.
CONCLUSION: MRI-based radiomics models using post-nCRT restaging images could predict residual EMVI and PLN metastasis in LARC. Incorporating mesorectal fat features improved model performance, suggesting that information from the surrounding mesorectal compartment may be useful for post-treatment risk assessment.