Deep learning-based ultrasound contrast tumor automatic identification and detection method

Abstract

The present invention discloses a deep learning-based ultrasound contrast tumor automatic identification and detection method. The method comprises the following steps of obtaining the characteristic video data of a target area; pre-processing the characteristic video data; partitioning the pre-processed characteristic video data to obtain a plurality of video blocks; inputting the video blocks in a 3D convolutional neural network to process the data and obtain a characteristic layer; inputting the characteristic layer in a full-connection layer to classify the characteristics, and finally determining whether the classification result is a tumor. By utilizing the 3D convolutional neural network, the method provided by the present invention can effectively extract the time-space effective blood flow characteristics in and between the ultrasound contrast video data frames, detects the tumor tissues based on the difference of the tumor tissues and the normal tissues and by combining the time-space characteristics of the video data and using a 3D CNN algorithm to segment the tumors, and has a better segmentation accuracy than a conventional machine learning algorithm.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a method for automatic recognition and detection of tumors in contrast-enhanced ultrasound based on deep learning. Background technique [0002] At present, there are few studies on computer-aided diagnosis methods for tumor diagnosis by contrast-enhanced ultrasound. By searching related keywords, it is found that there are few studies on this method in recent years. The main research idea is to use B-ultrasound images and known liver lesions. Images are used as data, and human neural networks and decision trees are used for classification, achieving a certain degree of accuracy. However, due to the diversity of lesion shapes, effective prediction cannot be achieved when dealing with unknown types of lesions. Among them, the recognition accuracy rate for liver cancer tissues with different tissue textures is only 65.2%, and the recognition rate for completely different...

AI Technical Summary

Problems solved by technology

However, due to the diversity of lesion shapes, it is impossible to achieve effective prediction when dealing with unknown types of lesions.

Among them, the recognition accuracy rate for liver cancer tissues with different tissue textures is only 65.2%, and the recognition rate for completely different types of liver cancer tissues is only 41.7%. Based on existing traditional classifiers such as decision trees, multi-classification logistic regression, etc., Segmentation accuracy is low, with high false positive rate and false negative rate; existing classifiers cannot achieve high accuracy in recognition

[0003] Ultrasound-enhanced tumor detection is one of the most sensitive and accurate detection methods in tumor detection. However, the existing ultrasound-enhanced tumor diagnosis is based on the professional knowledge of doctors and has high requirements for doctors' professional level. Therefore, there are only a few high-quality methods. Level hospitals can carry out ultrasound-enhanced tumor detection

Images