Research On Radius Rating Method Based On X-ray Images

Bone age is the important index in measuring physical development degree of children, which is widely used in clinical medicine, physical education, justice and other fields. As hand wrist bone can accurately reflect the whole body’s growth development status and is easy to take X-ray with low irradiation dose, it’s widely believed as the ideal body part to measuring the bone age by domestic and overseas scholars. CHINA 05 is a hand wrist bone maturity evaluation method currently used in our country. And the rating is the most critical steps of CHINA 05, which requires professionally trained staff who are proficient in bone development stages so that an effective manual judgment can be make. Additionally, the whole evaluation process is cumbersome and has heavy workload and takes long time. With the popularization of computer technology and the development of digital image processing techniques, realization of automatic rating currently has become a hot spot of bone age research. According to the CHINA 05, this paper propose a rating algorithm for radius, which has studied the automatic rating process of radius grade 8-0 to 8-4 in terms of the characteristics of radius within hand wrist X-ray image. The main contents are as following:(1) To propose a radius segmentation algorithm based on region growth method integrated with OTSU method. Firstly, to have gray transformation and image cutting for hand wrist bone X-ray image; Then, to block the cut image and to calculate the adaptive threshold value of each image block with OTSU method; Finally, To make similarity criterion combined with the adaptive threshold value and to realize effective segmentation for radius within hand wrist bone X-ray image by region growth method.(2) According to the evaluation criteria of radius grade 8-0 to 8-4 in CHINA 05 and the characteristics of radius within hand wrist X-ray image, we select partial representative gray feature and texture feature and propose fusion rate as a new feature at first, which is used the parameter of bone age feature. And then a classification experiment for radius grade was implemented with K Nearest Neighbor classifier and Random Forest classifier, respectively to validate the classification performance of the selected bone age features.(3) A radius rating system platform is established based on ITK、VTK and MFC. Through study on the radius rating process, the whole system was classified as five function modules including image reading, image display, image interaction, radius segmentation and automatic rating. Through analysis for the application technology of each module, the corresponding function was designed and realized.