AUTHOR=Xu Dong , Li Hao , Su Fanghui , Qiu Sizheng , Tong Huixia , Huang Meifeng , Yao Jianzhong 

TITLE=Identification of middle cerebral artery stenosis in transcranial Doppler using a modified VGG-16

JOURNAL=Frontiers in Neurology

VOLUME=Volume 15 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2024.1394435

DOI=10.3389/fneur.2024.1394435

ISSN=1664-2295

ABSTRACT=Objectives: The diagnosis of Intracranial atherosclerotic stenosis (ICAS) is of great significance for the prevention of stroke. Deep learning (DL) -based artificial intelligence techniques may aid in the diagnosis. The aim of this study was to identify ICAS in the middle cerebral artery based on a modified DL model. Methods: This retrospective study included two datasets. Dataset1 was 3068 TCD images of MCA from 1729 patients, which were judged as normal or stenosis by three physicians with different experiences combined with other medical imaging data, and used to improve and train the VGG16 models. Dataset2 was TCD images of 90 physical examination people, which were used to verify the robustness of the model and compare the consistency between the model and human physicians.The accuracy, precision, specificity, sensitivity and AUC of the best model VGG16+SE+SC on dataset1 reached 85.67±0.43(%),87.23±1.17(%),87.73± 1.47(%),83.60±1.60(%),0.857±0.004 and those on dataset2 were 93.70± 2.80(%),62.65±11.27(%),93.00±3.11(%),100.00±0.00(%),0.965±0.016.The kappa coefficient shows that it reaches the recognition level of senior doctors.The improved DL model has good diagnostic effect for MCV stenosis in TCD images, which is expected to provide help in ICAS screening.