Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT

D. Devakumar, Xin He, Eric Frey

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

The goal of this study was to compare two scatter compensation methods for myocardial perfusion SPECT imaging, namely the effective scatter source estimation (ESSE) and the dual energy window (DEW) methods. A population of 12 male Dynamic Cardiac Torso (DNCAT) anatomies were used. Low-noise projections were simulated for each organ of each anatomy using Monte Carlo simulation with scatter, attenuation and detector response modelling. The organ projections were then combined to form 36 sets of projections for each anatomy using randomly sampled activity ratios from a clinically realistic distribution. Poisson noise was then added to create clinically realistic projections with different noise levels for each phantom. Reconstruction was performed using OS-EM with attenuation and detector response compensation with the addition of no, ESSE or DEW scatter compensation. For EESE scatter compensation, the scatter was modelled for each iteration in the forward projector, and for DEW scatter compensation, the estimated scatter projections were obtained from a 5 keV wide energy window immediately below the primary photopeak window, scaled with a factor of 2.8 and filtered with a Gaussian filter. A channelized Hotelling observer was used to model human defect detection, and the area under the receiver operating characteristics curve (AUC) was used as a figure of merit We found that optimal cutoff frequency of the Gaussian filter applied to DEW estimate was 8 pixels FWHM. The AUC values for ESSE, DEW and no scatter compensation were 0.850, 0.843 and 0.839 respectively. The p value for ESSE vs. DEW was 0.0630, ESSE vs. no scatter compensation method was 0.0466 and DEW vs. no scatter compensation method was 0.1649. Both scatter compensation methods outperformed no scatter compensation and ESSE outperformed DEW, though only the difference between ESSE and no compensation was statistically significant at a confidence level of 5%.

Original languageEnglish (US)
Title of host publicationIEEE Nuclear Science Symposium Conference Record
Pages4472-4474
Number of pages3
Volume6
DOIs
StatePublished - 2007
Event2007 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS-MIC - Honolulu, HI, United States
Duration: Oct 27 2007Nov 3 2007

Other

Other2007 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS-MIC
CountryUnited States
CityHonolulu, HI
Period10/27/0711/3/07

Fingerprint

Compensation and Redress
Detectors
Cutoff frequency
Full width at half maximum
Pixels
Imaging techniques
Defect detection
Monte Carlo simulation

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Industrial and Manufacturing Engineering

Cite this

Devakumar, D., He, X., & Frey, E. (2007). Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT. In IEEE Nuclear Science Symposium Conference Record (Vol. 6, pp. 4472-4474). [4437104] https://doi.org/10.1109/NSSMIC.2007.4437104

Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT. / Devakumar, D.; He, Xin; Frey, Eric.

IEEE Nuclear Science Symposium Conference Record. Vol. 6 2007. p. 4472-4474 4437104.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Devakumar, D, He, X & Frey, E 2007, Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT. in IEEE Nuclear Science Symposium Conference Record. vol. 6, 4437104, pp. 4472-4474, 2007 IEEE Nuclear Science Symposium and Medical Imaging Conference, NSS-MIC, Honolulu, HI, United States, 10/27/07. https://doi.org/10.1109/NSSMIC.2007.4437104
Devakumar D, He X, Frey E. Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT. In IEEE Nuclear Science Symposium Conference Record. Vol. 6. 2007. p. 4472-4474. 4437104 https://doi.org/10.1109/NSSMIC.2007.4437104
Devakumar, D. ; He, Xin ; Frey, Eric. / Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT. IEEE Nuclear Science Symposium Conference Record. Vol. 6 2007. pp. 4472-4474
@inproceedings{379ba5449df044aca4f6ae2a59354ed0,
title = "Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT",
abstract = "The goal of this study was to compare two scatter compensation methods for myocardial perfusion SPECT imaging, namely the effective scatter source estimation (ESSE) and the dual energy window (DEW) methods. A population of 12 male Dynamic Cardiac Torso (DNCAT) anatomies were used. Low-noise projections were simulated for each organ of each anatomy using Monte Carlo simulation with scatter, attenuation and detector response modelling. The organ projections were then combined to form 36 sets of projections for each anatomy using randomly sampled activity ratios from a clinically realistic distribution. Poisson noise was then added to create clinically realistic projections with different noise levels for each phantom. Reconstruction was performed using OS-EM with attenuation and detector response compensation with the addition of no, ESSE or DEW scatter compensation. For EESE scatter compensation, the scatter was modelled for each iteration in the forward projector, and for DEW scatter compensation, the estimated scatter projections were obtained from a 5 keV wide energy window immediately below the primary photopeak window, scaled with a factor of 2.8 and filtered with a Gaussian filter. A channelized Hotelling observer was used to model human defect detection, and the area under the receiver operating characteristics curve (AUC) was used as a figure of merit We found that optimal cutoff frequency of the Gaussian filter applied to DEW estimate was 8 pixels FWHM. The AUC values for ESSE, DEW and no scatter compensation were 0.850, 0.843 and 0.839 respectively. The p value for ESSE vs. DEW was 0.0630, ESSE vs. no scatter compensation method was 0.0466 and DEW vs. no scatter compensation method was 0.1649. Both scatter compensation methods outperformed no scatter compensation and ESSE outperformed DEW, though only the difference between ESSE and no compensation was statistically significant at a confidence level of 5{\%}.",
author = "D. Devakumar and Xin He and Eric Frey",
year = "2007",
doi = "10.1109/NSSMIC.2007.4437104",
language = "English (US)",
isbn = "1424409233",
volume = "6",
pages = "4472--4474",
booktitle = "IEEE Nuclear Science Symposium Conference Record",

}

TY - GEN

T1 - Comparative assessment of ESSE and DEW scatter compensation methods for myocardial perfusion SPECT

AU - Devakumar, D.

AU - He, Xin

AU - Frey, Eric

PY - 2007

Y1 - 2007

N2 - The goal of this study was to compare two scatter compensation methods for myocardial perfusion SPECT imaging, namely the effective scatter source estimation (ESSE) and the dual energy window (DEW) methods. A population of 12 male Dynamic Cardiac Torso (DNCAT) anatomies were used. Low-noise projections were simulated for each organ of each anatomy using Monte Carlo simulation with scatter, attenuation and detector response modelling. The organ projections were then combined to form 36 sets of projections for each anatomy using randomly sampled activity ratios from a clinically realistic distribution. Poisson noise was then added to create clinically realistic projections with different noise levels for each phantom. Reconstruction was performed using OS-EM with attenuation and detector response compensation with the addition of no, ESSE or DEW scatter compensation. For EESE scatter compensation, the scatter was modelled for each iteration in the forward projector, and for DEW scatter compensation, the estimated scatter projections were obtained from a 5 keV wide energy window immediately below the primary photopeak window, scaled with a factor of 2.8 and filtered with a Gaussian filter. A channelized Hotelling observer was used to model human defect detection, and the area under the receiver operating characteristics curve (AUC) was used as a figure of merit We found that optimal cutoff frequency of the Gaussian filter applied to DEW estimate was 8 pixels FWHM. The AUC values for ESSE, DEW and no scatter compensation were 0.850, 0.843 and 0.839 respectively. The p value for ESSE vs. DEW was 0.0630, ESSE vs. no scatter compensation method was 0.0466 and DEW vs. no scatter compensation method was 0.1649. Both scatter compensation methods outperformed no scatter compensation and ESSE outperformed DEW, though only the difference between ESSE and no compensation was statistically significant at a confidence level of 5%.

AB - The goal of this study was to compare two scatter compensation methods for myocardial perfusion SPECT imaging, namely the effective scatter source estimation (ESSE) and the dual energy window (DEW) methods. A population of 12 male Dynamic Cardiac Torso (DNCAT) anatomies were used. Low-noise projections were simulated for each organ of each anatomy using Monte Carlo simulation with scatter, attenuation and detector response modelling. The organ projections were then combined to form 36 sets of projections for each anatomy using randomly sampled activity ratios from a clinically realistic distribution. Poisson noise was then added to create clinically realistic projections with different noise levels for each phantom. Reconstruction was performed using OS-EM with attenuation and detector response compensation with the addition of no, ESSE or DEW scatter compensation. For EESE scatter compensation, the scatter was modelled for each iteration in the forward projector, and for DEW scatter compensation, the estimated scatter projections were obtained from a 5 keV wide energy window immediately below the primary photopeak window, scaled with a factor of 2.8 and filtered with a Gaussian filter. A channelized Hotelling observer was used to model human defect detection, and the area under the receiver operating characteristics curve (AUC) was used as a figure of merit We found that optimal cutoff frequency of the Gaussian filter applied to DEW estimate was 8 pixels FWHM. The AUC values for ESSE, DEW and no scatter compensation were 0.850, 0.843 and 0.839 respectively. The p value for ESSE vs. DEW was 0.0630, ESSE vs. no scatter compensation method was 0.0466 and DEW vs. no scatter compensation method was 0.1649. Both scatter compensation methods outperformed no scatter compensation and ESSE outperformed DEW, though only the difference between ESSE and no compensation was statistically significant at a confidence level of 5%.

UR - http://www.scopus.com/inward/record.url?scp=48149104231&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=48149104231&partnerID=8YFLogxK

U2 - 10.1109/NSSMIC.2007.4437104

DO - 10.1109/NSSMIC.2007.4437104

M3 - Conference contribution

AN - SCOPUS:48149104231

SN - 1424409233

SN - 9781424409235

VL - 6

SP - 4472

EP - 4474

BT - IEEE Nuclear Science Symposium Conference Record

ER -