Research On Osteoporosis Detecting Based On Dental Panoramic Radiographs

As China’s aging society is coming, Osteoporosis disease gradually become one of the important factors of the old people. So that effective and simple osteoporosis detection methods cause strong interest of the scientific researchers. There is research shows that some feature on panoramic radiographs can describe whether a person is osteoporosis or normal, including cortical width, gray value, Fractal Dimension and soon.This paper’s main work based on how to build an effective and accurate method to detect osteoporosis. Paper’s content include data collection, feature extraction, pattern recognition and so on. This paper’s main content as follows:1) In order to identify mental foramen, I present a method based on OTSU and DLS, this method can identify mental foramen automatically.2) At CW feature extraction phrase, as Arifin’s method is not automatic, I present a new method based on cluster algorithm.3) Take FD value as another feature. When at feature extraction, do some improvement to traditional method Differential--Box-Counting method.4) At pattern recognition, I present a new method based on SVM, because the traditional method iterates repeated data too much. My method makes a threshold to the decision-making function, and add a limit of KKT which makes the algorithm keeps useful data,and remove the data with no help to SVM reconstruction.