Key point extraction and bone age prediction method in hip joint imaging

Abstract

The invention discloses a key point extraction and bone age prediction method in a hip joint image. The relevant key point extraction includes: using a feature extraction network to perform feature extraction on the hip joint image, and then obtaining the heat map of the C channel through down sampling; The feature extraction network is pre-trained with the hip joint images containing the positions of key points marked by experts; according to the size of the thermal value, N positions with the highest thermal value are detected from the heat map of the C channel as key points; among them, C and N are both natural numbers, and C is greater than N; by predicting the loss of key point deviation, the position deviation of N key points caused by the downsampling stage is compensated, and the final position of N key points is obtained. This method can accurately and quickly extract key points, greatly reduce the work intensity, and can realize bone age prediction.

Description

technical field [0001] The invention relates to the technical field of medical image analysis, in particular to a method for extracting key points in hip joint images and predicting bone age. Background technique [0002] At present, the analysis and processing of medical images mostly rely on manual methods. [0003] On the one hand, the accuracy of the analysis results completely depends on the technical level of the staff. Different staff may produce different analysis results for the same medical image. For some scientific research institutions, different analysis results or analysis whose accuracy is difficult to measure. The results will affect the follow-up scientific research work; at the same time, for some grassroots hospitals or related training structures, different analysis results or analysis results whose accuracy is difficult to measure will also have an adverse impact on the training objectives of new staff. [0004] On the other hand, due to the complete d...

AI Technical Summary

Problems solved by technology

[0003] On the one hand, the accuracy of the analysis results depends entirely on the technical level of the staff. Different staff may produce different analysis results for the same medical image. For some scientific research institutions, different analysis results or analysis whose accuracy is difficult to measure The results will affect the development of follow-up scientific research work; at the same time, for some grassroots hospitals or related training structures, different analysis results or analysis results whose accuracy is difficult to measure will also have a negative impact on the training objectives of new staff

[0004] On the other hand, due to the complete reliance on manual methods, the analysis efficiency is low and the work intensity is high

Images