Parkinsonism & related disorders · 2026
Deep learning-based MRI segmentation for substantia nigra in Parkinson's disease with cognitive impairment.
СтатьяНазвание 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.
Qi Weimin, Wang Jing, Yang Zixuan, Cheng Jiang, Niu Xiaoyan, Shao Na, Ren Yazhou, He Jianhang, Li Hui, Li Haining
Аннотация
BACKGROUND: Parkinson's disease (PD) is frequently accompanied by non-motor symptoms and cognitive impairment (PD-CI), highlighting the need for scalable imaging biomarkers for clinical stratification. Neuromelanin-sensitive MRI (NM-MRI) provides a pathology-adjacent measure of substantia nigra pars compacta (SNpc) degeneration, but automated SNpc quantification remains limited.
METHODS: We evaluated automated SNpc segmentation on NM-MRI using SA-U2Net and validated it against blinded manual delineation. The cohort included 53 PD-CI, 53 PD without cognitive impairment (PD-NCI), and 50 healthy controls (HC). SNpc area was quantified manually and automatically and compared across groups. Associations between bilateral SNpc area and clinical scales were screened using univariable analyses and tested using multivariable linear regression adjusting for age, sex, education, disease duration, and UPDRS-III, with false discovery rate (FDR) correction within subgroups.
RESULTS: SA-U2Net achieved accurate SNpc segmentation with good agreement to manual delineation. Both manual and automated measures showed significantly reduced SNpc area in PD-CI and PD-NCI compared with HC, with similar reductions between PD subtypes. In PD-CI, bilateral SNpc area remained independently and negatively associated with fatigue severity (FSS) after covariate adjustment and FDR correction. Other associations were attenuated after adjustment. In PD-NCI, no outcomes remained independently associated after covariate adjustment.
CONCLUSIONS: SA-U2Net enables reproducible automated SNpc quantification on NM-MRI and robust differentiation of PD from HC. Bilateral SNpc area shows a specific independent association with fatigue severity in PD-CI, while other associations require validation in multi-center longitudinal studies.
METHODS: We evaluated automated SNpc segmentation on NM-MRI using SA-U2Net and validated it against blinded manual delineation. The cohort included 53 PD-CI, 53 PD without cognitive impairment (PD-NCI), and 50 healthy controls (HC). SNpc area was quantified manually and automatically and compared across groups. Associations between bilateral SNpc area and clinical scales were screened using univariable analyses and tested using multivariable linear regression adjusting for age, sex, education, disease duration, and UPDRS-III, with false discovery rate (FDR) correction within subgroups.
RESULTS: SA-U2Net achieved accurate SNpc segmentation with good agreement to manual delineation. Both manual and automated measures showed significantly reduced SNpc area in PD-CI and PD-NCI compared with HC, with similar reductions between PD subtypes. In PD-CI, bilateral SNpc area remained independently and negatively associated with fatigue severity (FSS) after covariate adjustment and FDR correction. Other associations were attenuated after adjustment. In PD-NCI, no outcomes remained independently associated after covariate adjustment.
CONCLUSIONS: SA-U2Net enables reproducible automated SNpc quantification on NM-MRI and robust differentiation of PD from HC. Bilateral SNpc area shows a specific independent association with fatigue severity in PD-CI, while other associations require validation in multi-center longitudinal studies.