Rib small fracture detection method and system

Abstract

The invention provides a rib small fracture detection method and system. The method comprises the following steps: constructing a rib small fracture detection database; the database is a CT picture of which the fracture position is marked by a doctor; dividing the database into a training set, a verification set and a test set according to a preset proportion; preprocessing the CT picture; the preprocessing process comprises the following steps of: selecting three continuous CT images, sampling according to an HU value range, and performing weighted average to obtain three-channel images; inputting the three-channel image into an improved YOLOV5 model for training; and if the decrease of the continuous loss function is smaller than a preset threshold value, stopping storing the model parameters, and verifying whether the output result of the improved model is the same as the marked fracture position by using the test set. Based on the method, the invention further provides a rib small fracture detection system. A large amount of clinical data is used for training, the method has high generalization ability for data of different hospitals, and whether a patient has fine rib fractures or not can be rapidly and accurately detected.

Description

technical field [0001] The invention belongs to the technical field of rib detection, in particular to a method and system for detecting small rib fractures. Background technique [0002] With the progress of society and the development of science and technology, the application of medical imaging in disease diagnosis is becoming more and more extensive. As an important part of medical imaging, chest CT is also widely used in daily medical diagnosis. Rib fracture is a common chest injury, accounting for more than 85% of closed chest trauma. It is more common in middle-aged and elderly people, and less common in children. [0003] Because rib fractures are mostly caused by direct or indirect violence, a considerable part of which involves car accidents, industrial injuries, criminal cases and compensation issues, so missed diagnosis and misdiagnosis may cause medical disputes. However, due to the condition and position restrictions of emergency patients, hospital equipment ...

AI Technical Summary

Problems solved by technology

However, the execution efficiency of traditional machine learning algorithms is low and the results are not accurate enough, which cannot meet the needs of auxiliary diagnosis

Images