Segmentation And Growth Rate Prediction Of Central Lung Cancer Based On Petct Medical Image

Worldwide,lung cancer is one of the most common cancers and the malignant tumor with the highest mortality rate.This thesis of central type lung cancer tumor further study of PET/CT image processing algorithm,interested in the tumor area further explore accurate segmentation of the border,put forward a kind of PET/CT image sequences based rigid registration algorithm and the tumor lesion localization based on PET image segmentation algorithm,finally on PET/CT fusion image outline the tumor location.For image registration,because PET and CT sizes are different,double cubic interpolation is used to zoom in,and the contour curves of human body in PET and CT are extra CTed as the basis for registration.For tumor area segmentation,the boundary of the tumor in the PET images there is a transition of infiltration area,using the traditional image segmentation algorithm is difficult to achieve accurate segmentation,this paper used the SUVmax point as the center of the tumor area,take its dire CTion of eight pixels,and the two gradient algorithm,get eight dire CTion on the boundary of the optimal threshold,to take average as the final eight threshold image segmentation threshold.Finally,the results of image segmentation were evaluated according to the standard reference images manually sketched by two senior physicians in the nuclear medicine discipline.This paper proposes an automatic segmentation algorithm for central lung cancer tumors,which is helpful for the treatment evaluation and prognosis evaluation of cancer patients,improves the accuracy and efficiency of diagnosis and follow-up treatment,and reduces manual workload.