Medical image segmentation is widely used to identify and isolate specific areas of study within the image called the region of interest (ROI). The process of segmentation requires specialized and complex algorithms in order to implement computationally effective and accurate results. The objective of this thesis is to perform an experimental analysis of denoising and segmentation methods of medical images using parallel programming algorithms. These algorithms are based on the concept of cellular automata which can be effectively executed on graphic processing units (GPUs). The GPU's computing power offers promising performance on data intensive tasks due to their highly parallel architecture. The implementation of the algorithms are performed using an application programming interface (API) called Open Computing Language (OpenCL) that takes advantage of the parallelism of GPUs.
Date of publication
Thesis (Local Only Access)
Sanchez, Manuel, "Medical Image Segmentation using Cellular Automata: a GPU Case Study for the Efficient Implementation of a Denoising Algorithm and Seeded Speculation" (2014). Computer Science Theses. Paper 3.