METHOD AND APPARATUS FOR AUTOMATICALLY DETECTING FEATURE POINTS OF THREE-DIMENSIONAL MEDICAL IMAGE DATA BY USING DEEP LEARNING

摘要

A method for automatically detecting feature points of three-dimensional medical image data by using deep learning comprises the steps of: receiving an input of a three-dimensional volume medical image; generating a two-dimensional brightness value projection image on the basis of the three-dimensional volume medical image; automatically detecting initial anatomical feature points by using a first convolutional neural network, on the basis of the two-dimensional brightness value projection image; generating a three-dimensional volume-of-interest region on the basis of the initial anatomical feature points; and automatically detecting detailed anatomical feature points by using a second convolutional neural network, which is different from the first convolutional neural network, on the basis of the three-dimensional volume-of-interest region.