Optimum fourier filtering of cardiac data: A minimum-error method: Concise communication

S. L. Bacharach, M. V. Green, D. Vitale, G. White, M. A. Douglas, R. O. Bonow, S. M. Larson

Research output: Contribution to journalArticle

Abstract

Random fluctuations limit the accuracy of quantities derived from cardia time-activity curves (TACs). To overcome this problem, TACs are often fitted with a truncated Fourier series giveing rise to two sources of error: (a) the truncated series may not adequately describe the TAC shape, causing errors in parameters calculated from the fit: and (b) successive TACs acquired from the same subject under identical circumstances will fluctuate due to limited counts, causing the Fourier fits (and parameters derived from them) to fluctuate. These two errors, respectively, decrease and increase as the number of harmonics increases, suggesting the existence of a minimum in total error. This number of harmonics for minimum error (NHME) was calculated for each of six common parameters used to describe LV TACs. The 'true' value of each parameter was determined from TACs of very high stastical precision. Poisson noise was added to simulate lower count rates. For low-count TACs, used of either a smaller or a larger number of harmonics resulted in significantly greater error. NHME was found to occur at two harmonics for the systolic parameters studied, regardless of the noise level present in the TAC. For diastolic parameters, however, NHME was a strong function of the noise present in the TAC., varying from three harmonics for noise levels typical of regional TACs, to five or six harmonics for high-count global TACs.

Original languageEnglish
Pages (from-to)1176-1184
Number of pages9
JournalJournal of Nuclear Medicine
Volume24
Issue number12
Publication statusPublished - 1983

Fingerprint

Noise
Cardia
Fourier Analysis
Research Design

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology

Cite this

Bacharach, S. L., Green, M. V., Vitale, D., White, G., Douglas, M. A., Bonow, R. O., & Larson, S. M. (1983). Optimum fourier filtering of cardiac data: A minimum-error method: Concise communication. Journal of Nuclear Medicine, 24(12), 1176-1184.

Optimum fourier filtering of cardiac data : A minimum-error method: Concise communication. / Bacharach, S. L.; Green, M. V.; Vitale, D.; White, G.; Douglas, M. A.; Bonow, R. O.; Larson, S. M.

In: Journal of Nuclear Medicine, Vol. 24, No. 12, 1983, p. 1176-1184.

Research output: Contribution to journalArticle

Bacharach, SL, Green, MV, Vitale, D, White, G, Douglas, MA, Bonow, RO & Larson, SM 1983, 'Optimum fourier filtering of cardiac data: A minimum-error method: Concise communication', Journal of Nuclear Medicine, vol. 24, no. 12, pp. 1176-1184.
Bacharach SL, Green MV, Vitale D, White G, Douglas MA, Bonow RO et al. Optimum fourier filtering of cardiac data: A minimum-error method: Concise communication. Journal of Nuclear Medicine. 1983;24(12):1176-1184.
Bacharach, S. L. ; Green, M. V. ; Vitale, D. ; White, G. ; Douglas, M. A. ; Bonow, R. O. ; Larson, S. M. / Optimum fourier filtering of cardiac data : A minimum-error method: Concise communication. In: Journal of Nuclear Medicine. 1983 ; Vol. 24, No. 12. pp. 1176-1184.
@article{658ec3d2124241f2ab70f9aae0492a69,
title = "Optimum fourier filtering of cardiac data: A minimum-error method: Concise communication",
abstract = "Random fluctuations limit the accuracy of quantities derived from cardia time-activity curves (TACs). To overcome this problem, TACs are often fitted with a truncated Fourier series giveing rise to two sources of error: (a) the truncated series may not adequately describe the TAC shape, causing errors in parameters calculated from the fit: and (b) successive TACs acquired from the same subject under identical circumstances will fluctuate due to limited counts, causing the Fourier fits (and parameters derived from them) to fluctuate. These two errors, respectively, decrease and increase as the number of harmonics increases, suggesting the existence of a minimum in total error. This number of harmonics for minimum error (NHME) was calculated for each of six common parameters used to describe LV TACs. The 'true' value of each parameter was determined from TACs of very high stastical precision. Poisson noise was added to simulate lower count rates. For low-count TACs, used of either a smaller or a larger number of harmonics resulted in significantly greater error. NHME was found to occur at two harmonics for the systolic parameters studied, regardless of the noise level present in the TAC. For diastolic parameters, however, NHME was a strong function of the noise present in the TAC., varying from three harmonics for noise levels typical of regional TACs, to five or six harmonics for high-count global TACs.",
author = "Bacharach, {S. L.} and Green, {M. V.} and D. Vitale and G. White and Douglas, {M. A.} and Bonow, {R. O.} and Larson, {S. M.}",
year = "1983",
language = "English",
volume = "24",
pages = "1176--1184",
journal = "Journal of Nuclear Medicine",
issn = "0161-5505",
publisher = "Society of Nuclear Medicine Inc.",
number = "12",

}

TY - JOUR

T1 - Optimum fourier filtering of cardiac data

T2 - A minimum-error method: Concise communication

AU - Bacharach, S. L.

AU - Green, M. V.

AU - Vitale, D.

AU - White, G.

AU - Douglas, M. A.

AU - Bonow, R. O.

AU - Larson, S. M.

PY - 1983

Y1 - 1983

N2 - Random fluctuations limit the accuracy of quantities derived from cardia time-activity curves (TACs). To overcome this problem, TACs are often fitted with a truncated Fourier series giveing rise to two sources of error: (a) the truncated series may not adequately describe the TAC shape, causing errors in parameters calculated from the fit: and (b) successive TACs acquired from the same subject under identical circumstances will fluctuate due to limited counts, causing the Fourier fits (and parameters derived from them) to fluctuate. These two errors, respectively, decrease and increase as the number of harmonics increases, suggesting the existence of a minimum in total error. This number of harmonics for minimum error (NHME) was calculated for each of six common parameters used to describe LV TACs. The 'true' value of each parameter was determined from TACs of very high stastical precision. Poisson noise was added to simulate lower count rates. For low-count TACs, used of either a smaller or a larger number of harmonics resulted in significantly greater error. NHME was found to occur at two harmonics for the systolic parameters studied, regardless of the noise level present in the TAC. For diastolic parameters, however, NHME was a strong function of the noise present in the TAC., varying from three harmonics for noise levels typical of regional TACs, to five or six harmonics for high-count global TACs.

AB - Random fluctuations limit the accuracy of quantities derived from cardia time-activity curves (TACs). To overcome this problem, TACs are often fitted with a truncated Fourier series giveing rise to two sources of error: (a) the truncated series may not adequately describe the TAC shape, causing errors in parameters calculated from the fit: and (b) successive TACs acquired from the same subject under identical circumstances will fluctuate due to limited counts, causing the Fourier fits (and parameters derived from them) to fluctuate. These two errors, respectively, decrease and increase as the number of harmonics increases, suggesting the existence of a minimum in total error. This number of harmonics for minimum error (NHME) was calculated for each of six common parameters used to describe LV TACs. The 'true' value of each parameter was determined from TACs of very high stastical precision. Poisson noise was added to simulate lower count rates. For low-count TACs, used of either a smaller or a larger number of harmonics resulted in significantly greater error. NHME was found to occur at two harmonics for the systolic parameters studied, regardless of the noise level present in the TAC. For diastolic parameters, however, NHME was a strong function of the noise present in the TAC., varying from three harmonics for noise levels typical of regional TACs, to five or six harmonics for high-count global TACs.

UR - http://www.scopus.com/inward/record.url?scp=0021069214&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0021069214&partnerID=8YFLogxK

M3 - Article

C2 - 6644378

AN - SCOPUS:0021069214

VL - 24

SP - 1176

EP - 1184

JO - Journal of Nuclear Medicine

JF - Journal of Nuclear Medicine

SN - 0161-5505

IS - 12

ER -