Mammary gland X-ray image lesion detection method and device based on normal model learning

Abstract

The embodiment of the invention provides a mammary gland X-ray image lesion detection method and device based on normal model learning. The method comprises the following steps: segmenting a mammary gland region from a mammary gland X-ray image; extracting image blocks, and carrying out brightness normalization processing; selecting a part of normal region image blocks as a training set, and inputting the training set into the dual-depth convolutional neural network model for training to obtain a normal model; selecting a plurality of normal region image blocks from the training set as templates, and inputting the templates into the normal model to obtain feature vectors of template images; inputting the test set into a normal model to obtain a feature vector of the test image; and inputting the feature vectors of the template image and the test image into a nearest neighbor classifier to execute a class of classification to obtain a test result. The traditional lesion detection mode depending on lesion special image features is broken through, the problem that only specific types of lesions can be detected in the prior art is solved, all-around detection of all types of lesions isachieved, and the lesion detection rate and accuracy are comprehensively improved.

Description

technical field [0001] Embodiments of the present invention generally relate to the fields of image processing and deep convolutional neural networks, and more specifically, relate to a method and device for detecting mammogram lesions based on normal model learning. Background technique [0002] Breast cancer is currently the most common malignant tumor that endangers women's health around the world. In recent years, the incidence of breast cancer in Chinese women has increased significantly. Many regions in China have successively launched the "two cancers" screening work. Mammography is currently internationally recognized as an effective medical imaging method for breast cancer screening and early diagnosis, and plays a very important role in breast cancer screening. In recent years, with the rapid development of artificial intelligence technologies such as computer vision and machine learning, computer-aided diagnosis (Computer-Aided Diagnosis, CAD) technology has becom...

AI Technical Summary

Problems solved by technology

[0004] Although the existing mammography image lesion detection technology has achieved good detection results, these technologies are limited to the special image model of the lesion, can only detect specific types of lesions, have a single detection ability, and are not suitable for lesions outside the scope of the model description. Lack of detection capabilities, different detection techniques are required for different types of lesions

In addition, deep learning-based detection techniques face the overriding challenge of lacking a large-scale effective training set

In practice, the number of lesion samples is very limited, and the variety and difference are large, often lacking labels or annotations, which seriously hinders the effective lesion detection model through supervised learning, and the trained model is due to the lack of samples and imbalanced There is a high possibility of overfitting

[0005] In summary, the existing technology has low versatility, poor generalization ability, cannot detect multiple types of lesions at the same time, and faces problems such as lack of effective lesion training sets

Images