New fine-grained clustering algorithm on GPU architecture for bias field correction and MRI image segmentation
N. Aitali, Bouchaib Cherradi, Omar Bouattane, Mohamed Youssfi, Abdelhadi Raihani · 2015
In this paper, we propose a new fine-grained clustering bias field estimation and segmentation algorithm on Single Instruction Multiple Data (SIMD) architecture (GPU). The goal is to accelerate compute-intensive portions of the sequential version. We have implemented this parallel algorithm using Compute Unified Device Architecture (CUDA) on different NVidia GPU cards. The numerical results in terms of execution time show a gain up to 52x for GTX 580 versus the sequential implementation.