Optimization of compute unified device architecture for real-time ultrahigh-resolution optical coherence tomography

Ji Hyun Kim, Jaehong Aum, Jae Ho Han, Jichai Jeong

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

We propose an optimized signal processing scheme that utilizes the compute unified device architecture (CUDA) for real-time spectral domain optical coherence tomography (OCT). Because linear spline interpolation and the direct spectral reshaping method have low data and control dependencies, these algorithms maximally utilize graphic processing unit (GPU) resources for dispersion control. In addition, data transfer between main memory and GPU, regarded as one of the most wasteful and time-consuming processes in GPU computing, is executed in parallel with the signal processing by overlapping kernel execution and data transfers. Experimental results obtained from application of the proposed scheme to a laboratory constructed OCT system comprising five spectrally shifted SLDs indicate that the OCT system has an axial resolution of 4.8 μm and transverse resolution of 13 μm in air. Further, coherence artifacts are reduced by 3-14 dB over the side-lobes in the point spread function. The optimization of CUDA enables OCT imaging rates up to 350 kHz (A-lines/sec) with a single GTX680 GPU.

Original languageEnglish
Pages (from-to)308-313
Number of pages6
JournalOptics Communications
Volume334
DOIs
Publication statusPublished - 2015 Jan 15

Fingerprint

Optical tomography
tomography
optimization
Data transfer
Signal processing
signal processing
Optical transfer function
Splines
spectral methods
splines
point spread functions
Interpolation
lobes
interpolation
artifacts
resources
Imaging techniques
Data storage equipment
Graphics processing unit
Air

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Electrical and Electronic Engineering
  • Atomic and Molecular Physics, and Optics
  • Physical and Theoretical Chemistry

Cite this

Optimization of compute unified device architecture for real-time ultrahigh-resolution optical coherence tomography. / Kim, Ji Hyun; Aum, Jaehong; Han, Jae Ho; Jeong, Jichai.

In: Optics Communications, Vol. 334, 15.01.2015, p. 308-313.

Research output: Contribution to journalArticle

@article{7db43a67c3964d88a5cf2a4b86e05309,
title = "Optimization of compute unified device architecture for real-time ultrahigh-resolution optical coherence tomography",
abstract = "We propose an optimized signal processing scheme that utilizes the compute unified device architecture (CUDA) for real-time spectral domain optical coherence tomography (OCT). Because linear spline interpolation and the direct spectral reshaping method have low data and control dependencies, these algorithms maximally utilize graphic processing unit (GPU) resources for dispersion control. In addition, data transfer between main memory and GPU, regarded as one of the most wasteful and time-consuming processes in GPU computing, is executed in parallel with the signal processing by overlapping kernel execution and data transfers. Experimental results obtained from application of the proposed scheme to a laboratory constructed OCT system comprising five spectrally shifted SLDs indicate that the OCT system has an axial resolution of 4.8 μm and transverse resolution of 13 μm in air. Further, coherence artifacts are reduced by 3-14 dB over the side-lobes in the point spread function. The optimization of CUDA enables OCT imaging rates up to 350 kHz (A-lines/sec) with a single GTX680 GPU.",
keywords = "Near infrared optical reflectometry, Optical coherence tomography, Optical imaging, Parallel computing",
author = "Kim, {Ji Hyun} and Jaehong Aum and Han, {Jae Ho} and Jichai Jeong",
year = "2015",
month = "1",
day = "15",
doi = "10.1016/j.optcom.2014.08.067",
language = "English",
volume = "334",
pages = "308--313",
journal = "Optics Communications",
issn = "0030-4018",
publisher = "Elsevier",

}

TY - JOUR

T1 - Optimization of compute unified device architecture for real-time ultrahigh-resolution optical coherence tomography

AU - Kim, Ji Hyun

AU - Aum, Jaehong

AU - Han, Jae Ho

AU - Jeong, Jichai

PY - 2015/1/15

Y1 - 2015/1/15

N2 - We propose an optimized signal processing scheme that utilizes the compute unified device architecture (CUDA) for real-time spectral domain optical coherence tomography (OCT). Because linear spline interpolation and the direct spectral reshaping method have low data and control dependencies, these algorithms maximally utilize graphic processing unit (GPU) resources for dispersion control. In addition, data transfer between main memory and GPU, regarded as one of the most wasteful and time-consuming processes in GPU computing, is executed in parallel with the signal processing by overlapping kernel execution and data transfers. Experimental results obtained from application of the proposed scheme to a laboratory constructed OCT system comprising five spectrally shifted SLDs indicate that the OCT system has an axial resolution of 4.8 μm and transverse resolution of 13 μm in air. Further, coherence artifacts are reduced by 3-14 dB over the side-lobes in the point spread function. The optimization of CUDA enables OCT imaging rates up to 350 kHz (A-lines/sec) with a single GTX680 GPU.

AB - We propose an optimized signal processing scheme that utilizes the compute unified device architecture (CUDA) for real-time spectral domain optical coherence tomography (OCT). Because linear spline interpolation and the direct spectral reshaping method have low data and control dependencies, these algorithms maximally utilize graphic processing unit (GPU) resources for dispersion control. In addition, data transfer between main memory and GPU, regarded as one of the most wasteful and time-consuming processes in GPU computing, is executed in parallel with the signal processing by overlapping kernel execution and data transfers. Experimental results obtained from application of the proposed scheme to a laboratory constructed OCT system comprising five spectrally shifted SLDs indicate that the OCT system has an axial resolution of 4.8 μm and transverse resolution of 13 μm in air. Further, coherence artifacts are reduced by 3-14 dB over the side-lobes in the point spread function. The optimization of CUDA enables OCT imaging rates up to 350 kHz (A-lines/sec) with a single GTX680 GPU.

KW - Near infrared optical reflectometry

KW - Optical coherence tomography

KW - Optical imaging

KW - Parallel computing

UR - http://www.scopus.com/inward/record.url?scp=84907842338&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84907842338&partnerID=8YFLogxK

U2 - 10.1016/j.optcom.2014.08.067

DO - 10.1016/j.optcom.2014.08.067

M3 - Article

VL - 334

SP - 308

EP - 313

JO - Optics Communications

JF - Optics Communications

SN - 0030-4018

ER -