Ko Ji Su, Choi Yangsean, Jeong Eun Seon, Kim Hyun-Jung, Lee Grace Yoojin, Park Ji Eun, Kim Namkug, Kim Ho Sung
From the Department of Radiology and Research Institute of Radiology (J.S.K., Y.C., E.S.J., J.E.P., H.S.K.), University of Ulsan College of Medicine, Asan Medical Centre, Seoul, Republic of Korea.
From the Department of Radiology and Research Institute of Radiology (J.S.K., Y.C., E.S.J., J.E.P., H.S.K.), University of Ulsan College of Medicine, Asan Medical Centre, Seoul, Republic of Korea
AJNR Am J Neuroradiol. 2025 May 2;46(5):1007-1015. doi: 10.3174/ajnr.A8552.
The amount and distribution of cerebral microbleeds (CMB) are important risk factors for cognitive impairment. Our objective was to train and validate a deep learning (DL)-based segmentation model for cerebral microbleeds (CMBs) on SWI and to find associations among CMB, cognitive impairment, and vascular risk factors.
Participants in this single-institution retrospective study underwent brain MRI to evaluate cognitive impairment between January and September 2023. For training the DL model, the nnU-Net framework was used without modifications. The performance of the DL model was evaluated on independent internal and external validation data sets. Linear regression analysis was used to find associations among log-transformed CMB numbers, cognitive function (Mini-Mental Status Examination [MMSE]), white matter hyperintensity (WMH) burden, and clinical vascular risk factors (age, sex, hypertension, diabetes, lipid profiles, and body mass index).
Training of the DL model ( = 287) resulted in a robust segmentation performance with an average Dice score of 0.73 (95% CI, 0.67-0.79) in an internal validation set ( = 67) and modest performance in an external validation set (Dice score = 0.46; 95% CI, 0.33-0.59; = 68). In a temporally independent clinical data set ( = 448), older age, hypertension, and WMH burden were significantly associated with CMB numbers in all distributions (total, lobar, deep, and cerebellar; all < .01). The MMSE was significantly associated with hyperlipidemia (β = 1.88; 95% CI, 0.96-2.81; < .001), WMH burden (β = -0.17 per 1% WMH burden, 95% CI, -0.27-0.08; < . 001), and total CMB number (β = -0.01 per 1 CMB, 95% CI, -0.02-0.001; = .04) after adjusting for age and sex.
The DL model showed a robust segmentation performance for CMB. In all distributions, CMB had significant positive associations with WMH burden. Increased WMH burden and CMB numbers were associated with decreased cognitive function.
脑微出血（CMB）的数量和分布是认知障碍的重要危险因素。我们的目标是训练并验证基于深度学习（DL）的SWI序列脑微出血（CMB）分割模型，并找出CMB、认知障碍和血管危险因素之间的关联。
本单机构回顾性研究的参与者于2023年1月至9月接受脑部MRI检查以评估认知障碍。为训练DL模型，未作修改直接使用nnU-Net框架。在独立的内部和外部验证数据集上评估DL模型的性能。采用线性回归分析来找出对数转换后的CMB数量、认知功能（简易精神状态检查表[MMSE]）、白质高信号（WMH）负荷和临床血管危险因素（年龄、性别、高血压、糖尿病、血脂谱和体重指数）之间的关联。
DL模型在287例训练数据上，内部验证集（67例）的分割性能良好，平均Dice评分为0.73（95%CI，0.67 - 0.79），外部验证集的性能一般（Dice评分 = 0.46；95%CI，0.33 - 0.59；68例）。在一个时间独立的临床数据集（448例）中，年龄较大、高血压和WMH负荷与所有分布（总计、叶性、深部和小脑；均P <.01）的CMB数量显著相关。调整年龄和性别后，MMSE与高脂血症（β = 1.88；95%CI，0.96 - 2.81；P <.001）、WMH负荷（每1%WMH负荷β = -0.17，95%CI [-0.27, -0.08]；P <.001）和总CMB数量（每1个CMBβ = -0.01，95%CI [-0.02, -0.001]；P = 0.04）显著相关。
DL模型对CMB显示出良好的分割性能。在所有分布中，CMB与WMH负荷均有显著正相关。WMH负荷增加和CMB数量增多与认知功能下降有关。
