METHOD AND DEVICE FOR CONFIRMING DYNAMIC MULTIDIMENSIONAL LESION LOCATION BASED ON DEEP LEARNING IN MEDICAL IMAGE INFORMATION

摘要

Disclosed are a dynamic multidimensional lesion localization apparatus and method based on deep learning of medical image information. A dynamic multidimensional lesion localization method based on deep learning of medical image information according to an embodiment of the present invention includes: extracting an automated image including a lesion from an endoscopy image; generating a plurality of learning data for the automated image through pre-processing; creating n convolutional layers (wherein n is a natural number equal to or greater than 2) by filtering the plurality of training data; integrating the n convolutional layers into m fully connected layers (where m is a natural number equal to or greater than 2); and configuring a sequential combination of the n convolutional layers and the m fully connected layers to output an optimal structural layer including classification information for the lesion in the automated image.