Automatic CAC Voxel Classification with Multi-scale CNN Architecture

摘要: Coronary Artery Calcification (CAC) score is one of the most important measures in determining degree cardiovascular disease. It time-consuming to do this manually or semi-automatically, so automatic CAC scoring methods are being studied. Most classify calcified pixels(2D) voxels(2.5D 3D) and calculate score. We present a new voxel classification model with multi-scale CNN architecture which can reflect advantages large receptive small CNN. This study used cardiac CT dataset 98 patients from Asan Medical Center South Korea. The consisted DICOM raw data label annotated by radiologist. A total 10,000 voxels were selected for each artery(LM, CX, LAD, RCA) background(BG), 50,000 training testing. Our proposed showed an accuracy 89.58% Center. performance our generally better when compared other models.