Research On High Quality Reconstruction Algorithm For Four Dimensional Cone-beam CT

With the current incidence of cancer in society,people are increasingly concerned about the development of cancer prevention and control areas.Radiation therapy has become one of the main means of anti-tumor because of its noninvasive advantages.In recent years,with the development of related basic disciplines,scholars have combined radiotherapy and automatic control technology to develop three-dimensional conformal radiotherapy(3DCRT),Intensity Modulated Radiation Therapy,IMRT)and other technologies.These techniques are designed to reduce the size of the target area and made a large dose change in the target area internal and external at the same time.Therefore,these techniques require a accurate position in radiotherapy,the offset of the target position will lead to more serious side effects than traditional radiotherapy.Image-guided radiation therapy(IGRT)combines radiotherapy equipment with imaging equipment to perform real-time imaging during radiotherapy and adjust the target area to cover the tumor,dynamically.At present,the clinical IGRT system is commonly used in 3D-CBCT technology for imaging,scanning time is about one minute.Because of the presence of human respiratory movement,the collected CBCT data contains multiple respiratory cycles,making the reconstructed images with a large number of motion artifacts.This phenomenon is prominent in the patients with lung cancer.In order to solve the above problems,four-dimensional CBCT(4D-CBCT)imaging that incorporating time(phase)information on the basis of 3D-CBCT was proposed.Compared to 3D-CBCT imaging,4D-CBCT can provide the patient-specific respiratory motion information and multiple three-dimensional volumes to represent the different status in the breathing cycle.For the 4D-CBCT imaging,the respiratory signal would be recorded or estimated and CB projections were usually sorted into 8-10 subsets according to the respiratory signal.However,the gantry rotation speed and frame rate of the flat-panel imager limit the total number of cone-beam projections(i.e.,180-700 projections),which results in the relatively fewer projections at each respiratory phase.And the reconstructed CBCT images by using the conventional algorithm suffer from significant artifacts and noise.In addition,the randomness of breathing would lead to the cone-beam projections bunched into several clusters,and the bunched sampling scheme will aggravate the noises and artifacts level in the reconstructed images.In this paper,a large number of studies have been carried out with 4D-CBCT imaging technology,which aims to reconstruct high-quality 4D-CBCT sequence images by conventional CBCT projection data.The main work is as follows:1)we proposed to use the subset projections of the current phase and the image domain phase-correlated information for the 4D-CBCT reconstruction.But the very first point that needs to be made is that the image reconstructed from the full data set is no longer used as initialization or prior image.What we want to exploit is the phase-correlated redundant information,and motived by the success of block based image restoration,we proposed the Motion guided Spatiotemporal Sparsity(MgSS)approach for 4D-CBCT reconstruction.In this scheme,we try to divide the each phase CBCT images into cubes(three-dimensional blocks),track the cubes with estimated motion field vectors through time(phase),and then apply regional spatiotemporal sparsity on the tracked cubes.Specifically,we recast the tracked cubes into four-dimensional matrix,and use the higher order singular value decomposition(HOSVD)(Costantini et al.,2008)technique to analyze the regional spatiotemporal sparsity.And then the block based spatiotemporal sparsity was incorporated into a cost function for the reconstruction pass.One simple but effective optimization algorithm was used for the cost function solution.2)we are trying to explore the maximum temporal coherence of the spatial structure among 4D-CBCT phases.However,one key point is that when large motion exists,the difference of the two CBCT volumes at adjacent phases is not sparse enough to directly enforce temporal constraint.Thus,in this work,we conduct motion estimation/motion compensation(ME/MC)on the 4D-CBCT volume by using inter-phase deformation vector fields.Specifically,the optical flow based image registration approach is used to derive the DVFs.The motion compensated 4D-CBCT volume then can be viewed as a pseudostatic sequence which in combination with a spatio-temporal sparsifying transform,turns out to be highly sparse.Thus,in this work,we are encouraged to enforce the constraints on the motion-compensated 4D-CBCT volumes instead of the original data sets.The sparsifying transform used in this work is the 3D spatial total variation combined with 1D temporal total variation minimization.We subsequently construct a cost function for reconstruction pass,and minimizing this cost function corresponds to reconstructing each phase from a full set of projections.In this paper,simulation data experiment and patient real data experiment are used to verify the robustness of MgSS algorithm and Mc-tv algorithm.The experimental results show that the performance of both methods are better than traditional algorithms in 4D-CBCT image reconstruction.Both methods can reconstruct high quality 4D-CBCT images through the 4D-CBCT projection data,and can accurately and completely estimate the movement track and morphological changes of the lung tumor.It provides a basis for clinicians to adjust the radiotherapy plan in real time during radiotherapy.