A deep learning-based three-dimensional ultrasound abdominal wall hernia patch detection method and system

Abstract

The present invention provides a method and system for detecting a three-dimensional ultrasonic abdominal hernia patch based on deep learning, which relates to the field of image processing technology, and is characterized in that the method includes: collecting three-dimensional ultrasonic images through ABUS, extracting the coronal plane data of the image and then Carry out data amplification, perform patch calibration on the amplified image data and split to obtain the data set, build and train the convolutional neural network through the data set to output the prediction frame information, select the patch candidate area according to the prediction frame information, and then pass The NMS algorithm screens out the final patch detection results from the patch candidate regions. This method extracts the coronal surface and amplifies the original image data, which enhances the recognition accuracy of the neural network, improves the generalization ability and robustness of the neural network, and reduces the risk of fitting during network training. , adjust network structure parameters, accelerate network training and improve network performance, use NMS for clustering reconstruction to improve detection performance and accuracy.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a deep learning-based three-dimensional ultrasonic abdominal wall hernia patch detection method and system. Background technique [0002] With the rapid development of ultrasound imaging technology and patch materials, traditional ultrasound examinations such as hand-held ultrasound (HHUS) cannot fully and reliably identify today's popular lightweight (LW) patches, while automated three-dimensional breast ultrasound (ABUS) ), as a relatively mature innovative ultrasound imaging mode, compared with two-dimensional ultrasound, it can provide higher diagnostic accuracy, better prediction of lesion size, more intuitive visualization of lesion area and relationship between adjacent tissues, and has been successfully applied to The diagnosis of breast lesions and the diagnosis of abdominal wall hernia have also attracted the attention of scholars. However, the amount of data ...

AI Technical Summary

Problems solved by technology

However, the amount of data generated by ABUS for each patient is huge, and manual review of image data is extremely time-consuming and manpower-intensive, and the interpretation of images by different operators is different, and it is very easy to miss the diagnosis of weak abnormal areas

Images