Method and system for intelligent diagnosis of multiple model liver diffuse diseases based on ultrasound images

Abstract

An ultrasound image-based multi-model liver diffuse disease intelligent diagnosis method and system, the present invention relates to an intelligent diagnosis method for liver diffuse diseases by using depth learning algorithm to extract image features and image texture features and applying XGBoost algorithm to ultrasound images, It is the application of artificial intelligence method in ultrasonic image-based diagnosis, and can provide doctors with auxiliary suggestions for disease diagnosis. The invention comprises the following steps: 1, preprocessing the ultrasonic image of the liver; 2.preliminary classification of liver diffuse diseases based on convolution neural network; thirdly, the convolution neural network features and image texture features are combined to form multi-model features, and the XGBoost algorithm is used to achieve the final classification of liver diffuse diseases. The invention combines the depth learning algorithm and the traditional feature extraction algorithm, gives consideration to the shape feature and the texture feature of the image, and applies the XGBoost algorithm to improve the accuracy of the classification algorithm, and is suitable for the liver diffuse disease auxiliary diagnosis based on the ultrasonic image.

Description

(1) Technical field [0001] The invention belongs to the field of computer-aided diagnosis, and specifically relates to the diagnosis and research of diffuse liver diseases in ultrasonic images. It uses deep learning algorithms to extract image features and combine image texture features to obtain multi-model features, and applies XGBoost algorithm based on the multi-model features. A method for intelligent diagnosis of diffuse liver disease on ultrasound images. (2) Background technology [0002] Diffuse liver disease refers to diffuse pathological changes in the liver parenchyma, typically fatty liver and liver fibrosis. Liver histopathological examination is the gold standard for the diagnosis of fatty liver and liver fibrosis, but it is invasive and poor patient compliance. After diffuse liver lesions appear, the acoustic impedance of the lesion will also change, and this change will be reflected in the ultrasound image. Therefore, the sonographer can observe the change...

AI Technical Summary

Problems solved by technology

[0003] In the direction of intelligent diagnosis of diffuse liver diseases, the existing research at home and abroad is still based on traditional machine learning algorithms or signal processing algorithms. This method requires manual feature extraction, which is very time-consuming and laborious and requires professional knowledge. huge influence

Moreover, the manual features extracted based on small samples are application-specific and have low reproducibility. If the time-consuming and labor-intensive features designed are no longer applicable to a batch of images collected by other ultrasound equipment, it will be difficult to meet the clinical reality. demanded

Images