Support Vector Machine Applied Research In The Bone Age Assessment System

The skeletal age is a more credible criterion to assess individual growth because it reflects the level of growth and age characteristics. Assessment of skeletal maturity from radiographs of a child's hand-wrist is a new technology in recent thirty years. The main purpose of the assessment of skeletal age lies in discerning the skeleton development characteristic to obtain quantitative estimate. The tradition assessment of skeletal maturity involves comparison of a radiograph with textual and pictorial information from an atlas. That is a tedious and time-consuming process. Studying skeletal age, synthesizing assessment of individual growth, applying to evaluating and preventing teenagers' development state, discovering early and predicting disease, choosing talents in sports, etc. are the main point for choosing this. By using such practical knowledge as digital image processing, pattern recognition and computer vision techniques, it is quite useful for us to develop a computerized analysis system and automate estimating the skeletal age.So this paper focuses on studying the method of feature extraction and classifies the joint skeleton segmenting image. It puts forward a feature extraction algorithm method based on K-L transform and a method to assess skeletal age to the feature information extracted with the classification device based on SVM algorithm.It points out that feature extraction and feature selection is one of the key technologies in the image recognition system. According to the principle of feature extraction and selection, combining with the characteristics of radiograph, analyzing space distribution characteristics of wrist image, it puts forward the feature extraction algorithm and feature selection method based on K-L transform, and has done deep theory analysis and experiment to prove the feasibility of the algorithm. This method uses K-L transform to produce matrix of sample aggregate, get a low dimension sub space to describe the wrist bone's radiographs. Each image makes projection, and obtains one group of coordinate coefficients. The coefficients indicating the positionof this image in sub space can be discerned the image feature.Based on classification theory in the pattern-recognition, the paper puts forward a method to assess skeletal age to the feature information extracted with the classification device based on SVM algorithm. On the basis of existing sample datum, the paper has verified the feasibility and validity of feature extraction and classification algorithm of joint image of wrist bone, formed the frame of the assessment system of skeletal age.The paper makes some study in the fields that assess skeletal age with computer auxiliary, and has established the foundation for further developing and perfecting the bone age's computer automatic assessment system. It has certain reference function to study and solve other computer-assisted analysis of medical image.