Improving the Performance of Medical Image Processing Systems Using Graphics Processing Units (GPUs)

Abstract

Medical imaging currently plays a crucial role in all clinical applications of medical procedures, ranging from medical research to diagnosis and treatment planning. However, medical imaging procedures are often computationally demanding due to the large 3D medical datasets processed in practical clinical applications.

Thanks to the rapid enhanced performance of graphics processing units (GPUs), improved programming support, and their excellent cost-to-performance ratio, the GPU has emerged as a viable parallel computing platform for computationally intensive tasks requiring high processing power.

This study aims to leverage distributed computing techniques on the GPU to enhance and accelerate the processing of medical images, which could play a pivotal role in long-term therapeutic procedures. The algorithm was developed and applied in three key areas of medical image processing: segmentation, filtering, and edge detection.