Detection of Carotid Arteries in Magnetic Resonance Imaging Based on Deep Learning

Carotid atherosclerosis is the leading cause of death worldwide. Magnetic Resonance Imaging (MRI) techniques are commonly used to depict luminal stenosis resulting from atherosclerosis progression. This paper proposed a yolov3 based method to automatically detect carotid arteries in MRI images, which includes two branches: coordinate prediction and confidence prediction. The network also use the K-means clustering to get nine sizes of bounding box priors. Compared with other methods, this network has high accuracy and processing speed and can realize the automatic detection of carotid arteries, which greatly reduces the burden on doctors.