Study On Key Techniques Of Lung Tumor Radiotherapy Based On PET/CT Images

Lung cancer ranks first of malignant tumors threatening to human life and health, and radiotherapy is an important tool in clinical treatment of lung cancer. Computed tomography (CT) and positron emission tomography (PET) plays an important role in lung cancer radiotherapy. According to the theory and techniques of modern digital image processing, PET and CT images can help doctors determine the size of tumor and the position of radiation target, and assess and predict the tumor response to the current radiotherapy program, which allows doctors to work out and revise the treatment plan better. In this paper, image fusion, segmentation and tumor growth model which are the key technologies of lung cancer radiotherapy based on the PET/CT images are the main research content, and the main work is as follows:First of all, the paper proposes dual sparse dictionary based fusion algorithm for PET and CT images to break the limitation that the PET or CT alone is used for the assessment of treatment effect and the determination of tumor position. The algorithm takes the advantage that sparsity can well describe the image features, and takes the method of max selection with information of spatial domain as the fusion rule to fuse images, so the fused image can effectively retain the optimal information of PET and CT images which can help doctors locate radiation target. In addition, the algorithm are extended to the fusion of CT and MRI images of cerebral infarction and cerebral hemorrhage, and the fusion images are better than the other fusion algorithms which further proves the effectiveness of the algorithm.Secondly, in order to reduce the differences of manual tumor segmentation and improve the accuracy of segmentation, the paper proposes an automatic segmentation algorithm for lung tumors in PET images based on random walk method. The algorithm combines the threshold based method which is commonly used in the PET image segmentation and random walk method, and weighted gradient information is incorporated to the weighted boundary information, which solves the problem of the instability of segmentation results due to selecting different seeds, and improves the accuracy of segmentation. Then, in order to obtain more adaptive seeds, the paper proposed a segmentation method based on affinity propagation method and random walk method, which uses the advantage of automatic clustering and improves the accuracy of tumor segmentation further. The segmentation algorithm can assist in radiotherapy. Besides, volume-dose model is validated after obtaining the tumor volume by the segmentation method, and the result shows that the model can be well applied to the process of lung cancer radiotherapy which can provide guidance to doctors for revising the radiotherapy plan.Finally, in order to timely predict the effect on the tumor by the radiotherapy, the paper studies the tumor evolution model for individual patient based on PET images, which adds the treatment effect of the tumor. The use of the model can predict the spatial location of the tumor, helping doctors determine the next step treatment programs. The paper provides a number of rules in the parameter optimization process, which makes the model can forecast results more accurately and provides a certain reference value for the users of the model.