Találati lista | Tudóstér

001-es BibID:	BIBFORM015774
Első szerző:	Lévay István (fizikus)
Cím:	GPU technology in medical imaging / Lévay István, Koselák Mihály, Kis Sándor Attila, Opposits Gábor, Trón Lajos, Emri Miklós
Dátum:	2009
Megjegyzések:	Background: R&D projects by the Institute of Nuclear Medicine of the University of Debrecen comprise the development of PET-related data acquisition, image recon- struction, image processing software libraries, and test applications. An important task of this kind is to study how the 2D-3D image reconstruction, noise-reduction, and image registration methods can be implemented on HPC platforms. The appli- cation of a graphics processing unit as multiprocessor system is one of the most important HPC solutions, because GPU's computational performance is 2 orders of magnitude larger than that of CPU. The aim of the study was to develop special software components running on GPU hardware, significantly reducing the processing time of common 2D?3D iterative image reconstruction, 3D image filtering, manipulation, and image registration al- gorithms. Results: The developed components are part of a larger multimodal imaging pack- age developed by our institute. The easy to use and well scalable nVIDIA's hard- ware and CUDA software architecture have been used, because they have remark- ably good price/value ratio. Our results affect three image processing areas. The new PET-reconstruction component allows a reduction by factor of 8 the recon- struction runtime as compared to the conventional, CPU-based solutions. The run- ning time of GPU directed image manipulation and most of the 3D image filter algorithms were ~300 times faster than the same operations by an average com- puter, without GPU support. The cost-function calculation of the automatic image registration has been successfully improved.
Tárgyszavak:	Orvostudományok Elméleti orvostudományok idézhető absztrakt
Megjelenés:	Nuclear Medicine Review. - 12 : 1 (2009), p. 36. -
További szerzők:	Koselák Mihály (1986-) (informatikus) Kis Sándor Attila (1973-) (fizikus) Opposits Gábor (1974-) (fizikus, szoftver fejlesztő) Trón Lajos (1941-) (biofizikus) Emri Miklós (1962-) (fizikus)
Internet cím:	Szerző által megadott URL Intézményi repozitóriumban (DEA) tárolt változat
Borító:
Saját polcon: