Research And GPU Implemetation Of Raycasting Parallelized Algorithm On 3D Reconstruction

In recent years, the 3D medical image reconstruction has made significant progress. Ray-casting is the most widely used algorithm in 3D medical image reconstruction. Ray-casting algorithm can get more effective image information and have high value in clinical studies. However, it is difficult to achieve real-time display by using the traditional ray-casting algorithm because the reconstruction speed is limited by the large number of radiation sampling and integration operations.A novel parallel ray-casting which named Star-Raycasting is proposed In this dissertation, A novel operation is introduced and defined as a star operation, and star operations can be computed in parallel in the proposed algorithm compared with the serial chain of star operations in the Ray-Casting algorithm. The star operations meet associativity. The computation complexity of the proposed algorithm is reduced from O(n) to O(log2n)In this dissertation, a NVIDIA CUDA framework GPU is used to accelerate the reconstruction speed of ray-casting algorithm by multi-threaded parallel computing, which realizes fast real-time rendering. According to parallel architecture of CUDA, parallel resampling calculation is implemented by multi-threaded technology. At the same time, each casting ray is divided into multiple parts during merging stage through the star operation. Using the above approaches, the computational efficiency of ray-casting is improved.It is shown that, this modified method brings the computing time of Star-Raycasting algorithm down to 36.8 ms when reconstructing 2563 voxel, which increases an order of magnitude than that of traditional ray-casting algorithm. The experiment results show the effectiveness of the proposed Star-Raycasting.