Intelligent Bone Age Assessment System Of CHN Method Based On Feature Region Level Judgment

The bone ages of children and adolescents show their growth and development.Bone age assessment is widely used in medicine,sports,justice and other fields.The traditional clinical method of bone age assessment is to observe the bone maturity of multiple specific bones in the left hand whole-hand X-ray film by the doctors.Its accuracy depends on the subjective judgment ability of the doctors,and the assessment is time-consuming.The realization of automatic bone age assessment can reduce the work pressure of bone age doctors and improve the stability and reproducibility of bone age assessment.Traditional automatic bone age assessment uses image processing,appearance models,and other methods to extract features for recognition,and the evaluation results have large errors.At present,there are methods that use deep convolutional neural networks to automatically evaluate the bone age based on the whole palm bone image,but their accuracies are only slightly higher than the traditional method.In order to improve the accuracy and practicability of automatic bone age recognition,this thesis uses deep learning technology to develop an intelligent bone age assessment system based on the feature region level determination of the CHN method.The main research contents and results of this article are as follows:(1)In order to objectively evaluate the improvement of bone age assessment recognition rate brought by the method based on the classification of characteristic regions,firstly,an automatic bone age assessment method using the whole palm bone image which takes full advantage of depth learning technology is proposed.This method uses a simple image processing method to remove the image background noise and cut out the target area for training evaluation,without the need to perform bone level labeling,localization of regions of interest and so on.In addition,combined with the transfer learning method to improve the network performance,this method finally obtained the average bone age error of 0.539 year,and the accuracy rate of the error within 1.0 year reached 86.77%.(2)Based on the CHN bone age scoring method,14 characteristic region bones for bone age evaluation were segmented from each full palm bone,and then a corresponding convolutional neural network model is designed to evaluate the maturity of the characteristic area bones.In addition,considering that bone development is a continuous process,unlike traditional hard determination of bone maturity levels,the weighted score of bones is calculated using the classification probabilities of the two most probable levels output by the network.The experimental results of different network models show that the best result of this method is that the average bone age error is 0.415 year,and the accuracy rate of the error within 1.0 year reaches 95.48%,which is significantly better than the automatic bone age assessment method based on the whole wrist bone image.(3)The CHN method intelligent bone age assessment system based on the judgment of the characteristic region level is designed and developed.The system has the functions of expert assessment and audit,automatic bone age assessment,information record management,etc.It is an intelligent bone age assessment system for bone age experts.In this thesis,an intelligent bone age assessment method and system of CHN method is designed,which improves the accuracy of bone age assessment,and is of great significance to promote the practical application of automatic bone age assessment system.