Clinical imaging · 2026
Multi-scale DWI-based Radiomics for predicting infarct volume evolution in acute ischemic stroke beyond the therapeutic time window: A dual-center study.
СтатьяМногоцентровое исследованиеНазвание 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.
Liu Pei, Chen Yue, Liu Ziyu, Li Siyi, Zhu Xiaoling, Fan Xiaoli, Chen Yong
Аннотация
OBJECTIVES: This study examined whether radiomics features from multi-scale diffusion-weighted imaging (DWI) could predict infarct volume reduction during the subacute phase in acute ischemic stroke (AIS) patients beyond the therapeutic window, and whether predicted volume evolution correlates with clinical neurological function.
METHODS: Seventy-two AIS patients from Center 1 and 29 from Center 2 were retrospectively analyzed. Core infarct volumes of interest were delineated on baseline DWI, with volume reassessed on follow-up fluid-attenuated inversion recovery imaging. Four volumes of interest were studied: core, 2-mm peri-core, 5-mm peri-core, and merged. Radiomics features were extracted and reduced using least absolute shrinkage and selection operator regression. Models were built using logistic regression, XGBoost, LightGBM, and random forest with five-fold cross-validation in Center 1 and external validation in Center 2. Clinical data including neurological severity, stroke etiology, infarct laterality, and treatment regimen were collected and analyzed in relation to imaging outcomes.
RESULTS: The 2-mm peri-core logistic regression model performed best (cross-validated area under the curve [AUC] = 0.827 ± 0.064; external AUC = 0.804). A clinical model incorporating baseline infarct volume and treatment type achieved AUC = 0.726 ± 0.129, substantially below the radiomics model. NIHSS change correlated significantly with volume change (ρ = 0.291, p = 0.003). Neurological improvement was observed in 63.4% of the volume-improvement group versus 33.3% of the non-improvement group (p = 0.007).
CONCLUSIONS: Radiomics features from the 2-mm peri-core region provide robust prognostic value for infarct evolution in late-window AIS, with significant correlation with neurological function.
METHODS: Seventy-two AIS patients from Center 1 and 29 from Center 2 were retrospectively analyzed. Core infarct volumes of interest were delineated on baseline DWI, with volume reassessed on follow-up fluid-attenuated inversion recovery imaging. Four volumes of interest were studied: core, 2-mm peri-core, 5-mm peri-core, and merged. Radiomics features were extracted and reduced using least absolute shrinkage and selection operator regression. Models were built using logistic regression, XGBoost, LightGBM, and random forest with five-fold cross-validation in Center 1 and external validation in Center 2. Clinical data including neurological severity, stroke etiology, infarct laterality, and treatment regimen were collected and analyzed in relation to imaging outcomes.
RESULTS: The 2-mm peri-core logistic regression model performed best (cross-validated area under the curve [AUC] = 0.827 ± 0.064; external AUC = 0.804). A clinical model incorporating baseline infarct volume and treatment type achieved AUC = 0.726 ± 0.129, substantially below the radiomics model. NIHSS change correlated significantly with volume change (ρ = 0.291, p = 0.003). Neurological improvement was observed in 63.4% of the volume-improvement group versus 33.3% of the non-improvement group (p = 0.007).
CONCLUSIONS: Radiomics features from the 2-mm peri-core region provide robust prognostic value for infarct evolution in late-window AIS, with significant correlation with neurological function.