Application Research Of CBCT And DTS In Radiotherapy Positioning

The precision of patient setup is critical to the effect of radiation therapy. Factors that affecting setup errors include the operating skill of technicians, lose or gain weight of patient, variation of tumor's shape and size, position and relative movement of internal oragans, etc.To obtain anatomical information before implementing radation therapy, therefore, is a crucial step for precisely targeting the tumor. Image guided radiation therapy (IGRT) is a technic to minimize patient position uncertainty. During radiation therpy, three-dimensional anatomical information of the patient is acquired to correct set up errors, and hence, dose conformity is considerably enhanced as well as local control rate of tumor. In recent years, the Cone Beam CT (CBCT) which is based on electronic portal imaging devices (EPID) has made the daily use of portal imaging to adjust patient position possible. A series of 2D projection images are obtained in circular trajectory around the patient for CBCT images reconstruction or digital tomosynthesis (DTS) images reconstruction, the reconstructed CBCT volume is then registred with planning CT, or the DTS images are compared with reference DTS (RDTS) which is reconstructed by planning digitally reconstructed radiographs (DRRs), set up errors are calculated.This research reviews the application of CBCT and DTS in radiotherapy, and mainly focus on the study of CBCT and DTS images reconstruction.Rigid body registration algorithm based on multal information is also implemented for set up error correction. Main achievements are summarized as follow:(1)The FDK algorithm which is extensively applied in industry is adopted for CBCT reconstruction. Simulated phantom data, biological object projection data, as well as real patient projection images acquired by VARIAN 23EX linac are used for validating the algorithm. Satisfactory images have been reconstructed.(2) CBCT images are often plagued by ring artifacts which appear on the reconstructed images as a number of concentric rings superimposed on the scanned structures.The presence of these deleterious ring artifacts hampers post processing such as image registration, image segmentation, and implanted marker detection for patient positioning. A novel fast ring artifact reduction algorithm that can be applied directly on the reconstructed images is proposed.(3)Study and realize the mutual information based rigid body image registration algorithm, apply it to register the CBCT images and CT images.Experimental results show that using mutual information of pixel grey as optimization criterion can achieve precise image registration and sufficiently satisfy clinical needs.(4) FDK-alike algorithm is suggested and applied for DTS images reconstruction. Limited angles of projections are just needed for reconstruction. Experimental validation is implemented on the real patient projection images obtained by VARIAN 23EX linac.Definition of tissue and bone structure is greatly enhanced. This characteristic makes DTS an ideal alternative of set up error correction for moving organs or specific patients.(5)DRRs images reconstruction is realized by ray-tracing algorithm using planning CT images.The high quality image result demonstrates this algorithm's clinical potentials.