One aspect of the present disclosure can provide a cerebrovascular disease learning device. The cerebrovascular disease learning device may comprise: a maximum intensity projection magnetic resonance angiography (MIP MRA) construction unit which receives an input of 3D time-of-flight magnetic resonance angiography (3D TOF MRA) and constructs MIP MRA comprising a plurality of image frames; a spatial characteristic learning unit which constructs a spatial characteristic learning model based on the mode of convolutional neural network (CNN) learning, wherein the spatial characteristic learning model learns the spacial characteristics of each of the plurality of image frames; a frame characteristic learning unit which receives an input of the spatial characteristics and constructs a frame characteristic learning model based on the mode of recurrent neutral network (RNN) learning, wherein the frame characteristic learning model learns frame characteristics indicating the characteristics between the plurality of image frames; and a lesion characteristic learning unit which receives an input of the frame characteristics and constructs a lesion characteristic learning model based on the mode of CNN learning, wherein the lesion characteristic learning model learns the lesion characteristics of each of the plurality of image frames.
展开▼