MR Deep learning based prognosis classification via image feature of multimodal MR images

摘要

The present invention relates to a method for predicting patient prognosis using image features of deep learning-based multi-MR images, which configure a deep learning network and acquire an MR image for learning, and then apply the MR image for learning to the deep learning network with an image and repeatedly learning the correlation between image features; when the MR image of the patient is obtained, extracting image features of the MR image of the patient through the deep learning network; selecting a difference between the extracted image features by correcting the t-test and error detection ratio for each group; and obtaining and outputting patient prognostic information corresponding to the image feature based on the clustering method.