Benign and malignant lump identification system based on ultrasonic imaging

Abstract

The invention provides a benign and malignant lump identification system based on ultrasonic imaging. The system combines two ultrasonic modes of ultrasonic elastography and color Doppler ultrasound to carry out benign and malignant computer-aided identification on lumps, and comprises a preprocessing module, a feature extraction module, a classifier training module, a classifier fusion module andan aided identification module. According to the identification system provided by the invention, computer-aided identification is carried out by carrying out feature extraction, feature classification training and classification feature fusion on ultrasonic image data of lump focus elasticity and blood flow conditions; the identification system is high in lump feature classification precision, stable in performance and wide in application range, the lump structure can be analyzed more comprehensively, and therefore the accuracy of auxiliary diagnosis and recognition is improved.

Description

technical field [0001] The invention relates to the field of medical image classification, in particular to a system for identifying benign and malignant tumors based on ultrasound imaging. Background technique [0002] There are great differences in the treatment methods of benign and malignant masses. Before pathological examination, it is of great value to accurately diagnose the nature of masses through ultrasound imaging technology. Malignant masses are highly heterogeneous, and improving the diagnostic accuracy is directly related to the survival rate of patients. [0003] Ultrasound elastography (UE) displays the softness and hardness of human tissue as a color image, which is used to reflect the elasticity information of the lesion. Malignant lesions will lead to changes in tissue hardness. Therefore, UE can help doctors diagnose benign and malignant tumors, and can be used for clinical Diagnostics provide more accurate information. Color Doppler flow imaging (CDFI...

AI Technical Summary

Problems solved by technology

[0006] The invention patent with the application number CN201810052319.7 provides a multi-modal ultrasonic multi-level computer-aided diagnosis method for breast tumors, which extracts various characteristic parameters from multi-modal ultrasonic images, and uses the characteristic parameters after feature selection to train three Multi-layer breast tumor computer-aided diagnosis classifier, according to the multi-level computer-aided diagnosis classifier of multi-modal ultrasonic images, obtains multi-level auxiliary diagnostic information of whether breast tumors exist. This method uses various characteristic parameters of multi-modal ultrasonic images, which can Improve the performance of computer-aided diagnosis, but this method is only suitable for ultrasound computer-aided diagnosis of breast tumors, the scope of application is limited, and the classifier adopts a serial fusion method, connecting multiple classifiers in a head-to-tail manner, The output of the upper-level classifier is used as the input of the next-level classifier, and the accuracy of the final output diagnosis result is related to the accuracy of each classifier in the system

Images