Progressive multi-scale craniofacial bone fracture detection method

Abstract

The invention discloses a progressive multi-scale craniofacial bone fracture detection method, and belongs to the technical field of craniofacial bone fracture detection. The detection method comprises the following steps: firstly, cutting a fracture data set: cutting an original image with the center point of an original label as the center and with the length and width four times of the length and width of the original label as the length and width of a new image, and then carrying out primary training on the newly labeled data set, storing a trained pre-training model, and then training the original data set on the basis of the model. A heat map label is generated according to an original data label while training original data to extract features, then a region of interest is guided in a region generation network by comparing the heat map label, and finally, a candidate frame is continuously approaching to a real frame by narrowing a detection range. Compared with similar methods, the detection method has the advantages that a skull fracture part can be effectively detected, so that the fracture detection accuracy is improved, the probability of frame leakage is reduced, and the application range and the application scene are expanded.

Description

technical field [0001] The invention belongs to the technical field of craniofacial bone fracture detection, and in particular relates to a progressive multi-scale craniofacial bone fracture detection method. Background technique [0002] With the development of deep learning technology, target detection technology based on deep learning is gradually being applied more and more widely. Target detection is a target extraction technique that judges whether there are target instances of a given category (such as cars, cats, dogs) in the picture, and marks the location and category of each instance if it exists. The object detection method integrates various professional technologies such as deep learning, pattern recognition and digital image processing. There are three key points in target detection: (1) extraction of target features; (2) recognition of objects; (3) positioning of objects. Deep learning can realize image feature extraction and the implementation of target de...

AI Technical Summary

Problems solved by technology

Although the current deep learning network suitable for target detection has achieved very ideal results on public data sets, there are still problems of misjudgment and missed judgment for fracture data sets.

Images