3D Reconstruction Technology Of Brain MRI Images Based On Ray Casting Algorithm

In recent years,with the development of medical imaging technology,medical images have become an important basis for doctors to diagnose.However,the two-dimensional images provided by traditional imaging technology lack intuitiveness,and the clinical experience of doctors is mainly relied on to judge the three-dimensional shape of tissues and organs.The three-dimensional reconstruction technology of medical images can provide doctors with real three-dimensional models and assist doctors in diagnosis and treatment of diseases,which has important application value.The three-dimensional reconstruction technology of brain MRI images based on the ray casting algorithm is mainly studied in this paper.The main contributions of the thesis are as follows.Firstly,the method of acquiring MRI image and the characteristics of medical images are explained.The DICOM image file format is briefly analyzed.In order to obtain a higher quality model during subsequent 3D reconstruction,The original image is preprocessed,which mainly includes image filtering and segmentation.The image is smoothed by median filtering,and the main tissue structure of the brain is segmented by Otsu method.As a result,image data of brain tissue,skull and part of blood vessels are obtained.Secondly,the difference between the surface rendering algorithm and the volume rendering algorithm is compared.The moving cube algorithm in surface rendering is implemented,and the ambiguity problem in the algorithm and its solutions are introduced.The three-dimensional reconstruction of the segmented brain tissue,skull,blood vessel and other tissues is realized by C++ programming.And the results are integrated to obtain a complete brain model.Then,the thesis focuses on the optimization process of the ray casting algorithm in the volume rendering algorithm.The realization principle and key technology of the ray casting algorithm are studied,and the algorithm is optimized by the technology of invalid voxel elimination and early termination of ray synthesis on the CPU side.The algorithm is implemented by C++ programming,and the three-dimensional reconstruction of multiple sets of images is carried out.The algorithm before and after optimization is compared and analyzed.Finally,the acceleration algorithm of the ray casting algorithm based on CUDA technology has been implemented.Traditional ray casting algorithms are accelerated by the parallel computing capabilities of CUDA,and the amount of calculation during ray casting is reduced from the algorithm level,thereby further improving the efficiency of the algorithm.The algorithm is programmed to realize the three-dimensional reconstruction of multiple sets of images.The execution efficiency of this algorithm is compared with the traditional ray casting algorithm and the ray casting algorithm in the VTK library.