RGBA packing for fast cone beam reconstruction on the GPU

Fumihiko Ino; Seiji Yoshida; Kenichi Hagihara

doi:10.1117/12.811149

13 March 2009 RGBA packing for fast cone beam reconstruction on the GPU

Fumihiko Ino, Seiji Yoshida, Kenichi Hagihara

Proceedings Volume 7258, Medical Imaging 2009: Physics of Medical Imaging; 725858 (2009) https://doi.org/10.1117/12.811149
Event: SPIE Medical Imaging, 2009, Lake Buena Vista (Orlando Area), Florida, United States

Abstract

This paper presents a fast cone beam reconstruction method accelerated on the graphics processing unit (GPU). We implement the Feldkamp, Davis, and Kress (FDK) algorithm on the OpenGL graphics pipeline, which allows us to exploit the full resources and capabilities available on the GPU. The proposed method differs from previous GPU-based methods in having an RGBA packing scheme capable of directly dealing with projections without rebinning. It also reduces the amount of computation by using a data reuse scheme, which is useful to save the memory bandwidth for this memory-intensive problem. Both schemes contribute to reduce the number of rendering passes, namely the number of kernel invocations on the GPU, realizing fast cone beam reconstruction. We show some experimental results obtained on a desktop PC with an nVIDIA GeForce 8800 GTX card. As a result, the proposed method takes 8.1 seconds to reconstruct a 512³-voxel volume from 360 512²-pixel projection images. This execution time is equivalent to a 15.6-fold speedup over a CPU implementation, showing 10% higher performance as compared with a previous OpenGL-based method that requires the single-slice rebinning of projections for acceleration. With respect to non-rebinned data, our method provides approximately three times higher performance than the previous method.

Citation Download Citation

Fumihiko Ino, Seiji Yoshida, and Kenichi Hagihara "RGBA packing for fast cone beam reconstruction on the GPU", Proc. SPIE 7258, Medical Imaging 2009: Physics of Medical Imaging, 725858 (13 March 2009); https://doi.org/10.1117/12.811149

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available