| Summary: | 碩士 === 華梵大學 === 資訊管理學系碩士班 === 97 === This paper is to develop a system that could automatically compute the cobb’s angle in digital radiographic images to estimate the severity and the risk of Scoliosis. In this thesis, we find the necessary information about the spine section and then compute the cobb’s angle. At present, we use the edge of the spine model to generate the templates and do matching by templates.
In the template matching, there are two methods: edge detection and distance transform. We detect the vertebrae by templates on the radiographic image and find necessary information about the spine section.
In order to improve the speed and accuracy, we remove most of the unnecessary information according to brightness, and use the histogram to select the region of interest, before the template matching. Then, we get the necessary information about the spine section, and then, we use the cobb angle measurement to compute the cobb’s angle.
There are 16 images for testing the accuracy. The accuracy is 50% in 10 degree error. And, the accuracy is 18.75% in 5 degree error. The average error is 14.25 degree. The standard deviation error is 10.89.
|
|---|
