Details

Autor(en) / Beteiligte

• Hsin-Yi Tsai
• Zhang Hanyu
• Che-Lun Hung
• Hsian-Min Chen

Titel

GPU-based Gray-Level Co-occurrence Matrix for Extracting Features from Magnetic Resonance Images

Ist Teil von

• 2017 14th International Symposium on Pervasive Systems, Algorithms and Networks & 2017 11th International Conference on Frontier of Computer Science and Technology & 2017 Third International Symposium of Creative Computing (ISPAN-FCST-ISCC), 2017, p.391-396

Ort / Verlag

IEEE

Erscheinungsjahr

2017

Link zum Volltext

• IEEE_Xplore

Quelle

IEEE/IET Electronic Library (IEL)

Beschreibungen/Notizen

• With the continuously increasing power of computation, especially in the region of parallel computing, computerbased texture analysis, computer-assisted classification methods, automated pathology detections, etc. are more and more commonly performed on medical images, like X-ray, Magnetic Resonance (MR) images, for clinical or scientific purposes. These procedures almost always include a stage of textural feature extraction, which usually requires an extensive computation. In this paper, we propose a GPGPU (General-purpose computing on graphics processing units)-based parallel method to accelerate the extraction of a set of features based on the Gray-Level Co-Occurrence Matrix (GLCM) which is a second order statistic that characterizes textures. Performance evaluation of the proposed method implemented with CUDA C is carried out on various GPU devices by comparing to its serial counterpart which is implemented in both Matlab and C on a single node. A series of experimental tests focused on Magnetic Resonance (MR) brain images demonstrate that the proposed method is very efficient and superior to the serial counterpart. A speedup of about 30 - 100 fold is achieved in general.