Research On Segmentation And Measurement Of Tumor In CT Image

Measurement of tumor is a significant index in clinical diagnosis and radiotherapy. If tumor can be segmented and measured correctly, the accuracy and effectiveness of diagnosis will be improved largely.The measurement of tumor is based on the segmentation of tumor, and only when the tumor is segmented correctly, the measurement can make sense. Nowadays the segmentation of medical image is still the hot point of research.The content of this thesis includes some parts as follows:A hybrid segmentation method is proposed for the purpose of exploiting the particular advantages of traditional seeded region growing (SRG) and mathematical morphology, which can solve the problem of lung tumor adjacent to tissue of similar attenuation. This method also can be applied to the tumor segmentation of other tissues in CT image.The tumor area is calculated by polygonal approximation, some data of area are approximated by Tchebyshev multinomial algorithm to get the expression about the change of area, and then the volume of tumor is obtained by integral in the CT scope which contains the tumor. This method can get higher precision of tumor volume with the lower data quantity.The experiment shows that the method of segmentation and measurement of tumor proposed in this paper has certain reliability in clinical diagnosis.