Semantic enhanced hash medical image retrieval method based on mixed attention

Abstract

The invention relates to a semantic enhanced hash medical image retrieval method based on mixed attention. Firstly, a data set is divided into a training set and a test retrieval set, images are randomly selected from the training set to form a medical triple, then an overall network model is constructed, a medical triple sample serves as input of the network model, finally, the overall network model is trained, and a retrieval result is obtained through a trained network. A channel attention module and a space attention module are utilized to form a mixed attention mechanism, and ROI information can be efficiently extracted; the learning process of the Hash codes is restrained by utilizing category-level semantic information, so that different categories of similar Hash codes can be distinguished; and when the depth embedding is mapped to the discrete hash code, the quantization error between the depth embedding and the hash code is reduced by using the quantization loss item, so that the precision of medical image retrieval can be further improved.

Description

technical field [0001] The invention belongs to the field of medical image retrieval, in particular to a semantically enhanced hash medical image retrieval method based on mixed attention. Background technique [0002] With the rapid development of radiography technology, medical data is gradually digitized, and the number of medical images has increased dramatically. In order to better assist medical diagnosis and evaluation, it is critical to mine useful information in large-scale medical images. Therefore, medical image retrieval has attracted extensive attention. [0003] Medical image retrieval can be divided into two categories: text-based medical image retrieval and content-based medical image retrieval. Text-based medical image retrieval appeared in the early days of medical image retrieval. It avoids the analysis of medical image visualization elements, indexes medical images from the aspects of name, size, type, etc., and often queries medical images based on key...

AI Technical Summary

Problems solved by technology

However, most of the existing content-based medical image retrieval methods only learn the relative relationship of medical images to extract deep features, while ignoring the category-level semantics of medical images and labels, resulting in the problem of insufficient utilization of high-level semantic information, which ultimately affects retrieval performance

Images