GPU Accelerated 3D Image Deformation Using Thin-Plate Splines

3D image deformation for medical image registration generally is a time-consuming task. This drawback slows down the speed of image registration. Thin-Plate Spline (TPS) is a commonly used interpolation technique to deform images which assures the least bending energy. This paper proposes a parallel implementation of 3D image deformation using Thin-Plate Splines and tri-linear interpolation which are based on CPU + GPU heterogeneous platform. We address the computational model including thread partition and memory allocation to analyze the performance of parallel algorithm on GPU. Using CUDA C and NIVIDA Tesla C2050 with 448 stream proceesors threads, we achieve an approximately 70-fold increase in speed in 3D medical image deformation, which shows higher speed than CPU on the final result. Experiments show that this GPU computational model is a practical way to accelerate image deformation.