Thyroid tumor image classification method based on multiple modes and terminal equipment

Abstract

The invention discloses a thyroid pathology image classification method and terminal equipment based on multiple modalities, and the method comprises the steps: carrying out the information feature extraction of thyroid pathology images of three modalities through employing three ResNet18 networks, obtaining three-modal information features, and carrying out the classification of thyroid pathology images of three modalities; the thyroid pathology images in the three modes comprise a thyroid ultrasound image, a thyroid elasticity image and a thyroid blood flow image; adopting a multi-mode multi-head attention module to extract common information features of the thyroid pathology images of the three modes; and fusing the three-mode information features and the common information features, performing thyroid pathology image classification by using a residual network, and outputting a classification result. The designed multi-modal thyroid pathology image classification method is verified on a multi-modal thyroid ultrasound data set provided by a cooperative research unit, and a result proves that the thyroid pathology images can be accurately classified by the method, so that rapid and accurate assistance is provided for diagnosis of thyroid cancer by an ultrasound department doctor.

Description

technical field [0001] The present invention relates to the technical field of image recognition, in particular to a multimodal-based thyroid tumor image classification method and terminal equipment. Background technique [0002] Thyroid cancer is the most common thyroid malignancy, accounting for about 1% of systemic malignancies, and can occur in multiple age groups. The structure of thyroid tissue is complex, and there are many interference factors in ultrasound images. Therefore, it is difficult for physicians to directly classify lesion features in ultrasound images of thyroid cancer. In order to alleviate this situation, in addition to thyroid ultrasound imaging, doctors usually combine ultrasound elastography and ultrasound blood flow imaging to more accurately identify thyroid cancer images. [0003] Currently, deep learning has achieved promising results in computer-aided diagnosis due to the advantage of obtaining high-dimensional features. However, most of the ...

AI Technical Summary

Problems solved by technology

[0005] In view of the above deficiencies in the prior art, the purpose of the present invention is to provide a multimodal-based thyroid tumor image classification method and terminal equipment, aiming to solve the problem of low classification accuracy of thyroid cancer images in the prior art

Images