It has been suggested that the association between white matter hyperintensity (WMH) and carotid arterial features includes carotid plaques and carotid intima-media thickness, arterial stiffness, or remodeling. Under the hypothesis that white, matter ...
It has been suggested that the association between white matter hyperintensity (WMH) and carotid arterial features includes carotid plaques and carotid intima-media thickness, arterial stiffness, or remodeling. Under the hypothesis that white, matter abnormalities correlate with the configuration of the feeding carotid arteries, we use deep learning to estimate WMH volumes based solely on time-of-flight (TOF) carotid MRA images. To improve performance, we use transfer learning from a pre-trained V-Net for carotid artery segmentation. We evaluate the WMH volume prediction performance by applying Grad-CAM and confirm that our proposed transfer learning model estimates the WMH volume derived from carotid arterial features.