Accelerating compute intensive medical imaging segmentation algorithms using hybrid CPU-GPU implementations

• Published in: Multimedia tools and applications 2017-02, Vol.76 (3), p.3537-3555
• Main Authors: Alsmirat, Mohammad A., Jararweh, Yaser, Al-Ayyoub, Mahmoud, Shehab, Mohammed A., Gupta, Brij B.
• Format: Article
• Language: English
• Subjects: Accuracy; Algorithms; Clustering; Computer Communication Networks; Computer Science; Data Structures and Information Theory; Fuzzy; Image processing; Image processing systems; Image segmentation; Information technology; Medical imaging; Multimedia; Multimedia Information Systems; Special Purpose and Application-Based Systems; Studies
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-016-3884-2

Medical image processing is one of the most famous image processing fields in this era. This fame comes because of the big revolution in information technology that is used to diagnose many illnesses and saves patients lives. There are many image processing techniques used in this field, such as image reconstructing, image segmentation and many more. Image segmentation is a mandatory step in many image processing based diagnosis procedures. Many segmentation algorithms use clustering approach. In this paper, we focus on Fuzzy C-Means based segmentation algorithms because of the segmentation accuracy they provide. In many cases, these algorithms need long execution times. In this paper, we accelerate the execution time of these algorithms using Graphics Process Unit (GPU) capabilities. We achieve performance enhancement by up to 8.9x without compromising the segmentation accuracy.