G C Sharp et al 2007 Phys. Med. Biol. 52 5771 doi:10.1088/0031-9155/52/19/003
G C Sharp1, N Kandasamy2, H Singh2 and M Folkert1
Show affiliationsThis paper shows how to significantly accelerate cone-beam CT reconstruction and 3D deformable image registration using the stream-processing model. We describe data-parallel designs for the Feldkamp, Davis and Kress (FDK) reconstruction algorithm, and the demons deformable registration algorithm, suitable for use on a commodity graphics processing unit. The streaming versions of these algorithms are implemented using the Brook programming environment and executed on an NVidia 8800 GPU. Performance results using CT data of a preserved swine lung indicate that the GPU-based implementations of the FDK and demons algorithms achieve a substantial speedup—up to 80 times for FDK and 70 times for demons when compared to an optimized reference implementation on a 2.8 GHz Intel processor. In addition, the accuracy of the GPU-based implementations was found to be excellent. Compared with CPU-based implementations, the RMS differences were less than 0.1 Hounsfield unit for reconstruction and less than 0.1 mm for deformable registration.
07.05.Bx Computer systems: hardware, operating systems, computer languages, and utilities
Issue 19 (21 September 2007)
Received 12 June 2007, in final form 23 July 2007
Published 10 September 2007
G C Sharp et al 2007 Phys. Med. Biol. 52 5771
Adam de la Zerda et al 2007 Phys. Med. Biol. 52 4137
Claude Carignan et al. 1998 The Astronomical Journal 116 1690
Fang Xu and Klaus Mueller 2007 Phys. Med. Biol. 52 3405
Claire McCann and Michael D Sherar 2006 Phys. Med. Biol. 51 3835
E P O'Reilly et al 2004 J. Phys.: Condens. Matter 16 S3257
G J Price et al 2009 Phys. Med. Biol. 54 6515
W Li et al 2000 J. Phys.: Condens. Matter 12 L269
J M Warman et al 2009 Phys. Med. Biol. 54 3185
Chunhua Men et al 2009 Phys. Med. Biol. 54 6565