Original contribution
Accuracy of gamma-variate fits to concentration-time curves from dynamic susceptibility-contrast enhanced MRI: Influence of time resolution, maximal signal drop and signal-to-noise

https://doi.org/10.1016/S0730-725X(96)00392-XGet rights and content

Abstract

Concentration-time curves derived from dynamic susceptibility-contrast enhanced magnetic resonance imaging are widely used to calculate cerebrovascular parameters. To exclude effects of recirculation, a nonlinear regression method is used to fit a Γ-variate function to the concentration-time course. In previous studies the errors arising from the fitting procedure have not been quantified. In a computer simulation we investigate the uncertainties of parameters calculated from the fitted Γ-variate function, exploring the dependencies on signal-to-noise (SNR), time resolution (Δt), and maximal signal drop (MSD). Our study was performed to give a framework on how to design MR-sequences and choose contrast media and their application in order to yield concentration-time curves which allow a reliable performance of the Γ-variate fitting procedure. We recorded 396 concentration-time curves from regions of interest of 40 patients. The Γ-variate fitting procedure was applied to these curves resulting in 396 parameter sets. Ideal concentration-time curves as Γ-variate functions were generated from these sets with a given Δt, MSD, and SNR. Recirculation effect was simulated. Then the Γ-variate fitting was performed again. From ideal and simulated Γ-variate function the area and the normalized first moment were calculated. The uncertainties of the values calculated from the simulated curve relating to the values of the original one were determined. Increase of SNR decreases the involved errors. With SNR values of 100 and more there is only minor influence of Δt and MSD and the fitted curve approximates the original data very well. Smaller values of SNR lead to a stronger influence of Δt and MSD and a higher number of fitting failures. With increasing Δt the uncertainties also increase. Intermediate values of MSD (30% to 70%) yield the smallest errors while increasing or decreasing MSD yields an increase of uncertainty. To achieve low uncertainties in the calculation of cerebrovascular parameters from Γ-variate fits, Δt of the imaging sequence and MSD must be considered. This is more important the lower SNR is. The shown depenencies should be taken into account when choosing MR sequence parameters and application of contrast media.

References (28)

  • J.W. Belliveau et al.

    Functional cerebral imaging by susceptibility-contrast NMR

    Magn. Reson. Med.

    (1990)
  • B.R. Rosen et al.

    Perfusion imaging with NMR contrast agents

    Magn. Reson. Med.

    (1990)
  • B.R. Rosen et al.

    Contrast agents and cerebral hemodynamics

    Magn. Reson. Med.

    (1991)
  • P. Meier et al.

    On the theory of the indicatordilution method for measurement of blood flow and volume

    J. Appl. Physiol.

    (1954)
  • L. Axel

    Cerebral blood flow determination by rapid-sequence computed tomography: A theoretical analysis

    Radiology

    (1980)
  • R.R. Edelman et al.

    Cerebral blood flow: Assessment with dynamic contrast-enhanced T2-weighted MR imaging at 1.5T

    Radiology

    (1990)
  • H. Aaronen et al.

    Susceptibility-contrast CBV imaging: Optimization of contrast dose and imaging sequences

  • R.M. Weisskoff et al.

    Pitfalls in MR measurement of tissue blood flow with intravascular tracers: Which mean transit time?

    Magn. Reson. Med.

    (1993)
  • D.W. Marquardt

    An algorithm for least squares estimation of non-linear parameters

    J. Soc. Ind. Appl. Math.

    (1963)
  • P.R. Bevington
  • W.H. Press et al.
  • M.D. Robson et al.

    Errors associated with fitting dynamic susceptibility contrast images for cerebral perfusion calculations

  • R.H. Huesman et al.

    Kinetic data analysis with a noisy input function

    Phys. Med. Biol.

    (1987)
  • D. Lu et al.

    Effect of sample numbers on the kinetic data analysis of MR contrast agents

    Magn. Reson. Med.

    (1993)
  • Cited by (0)

    View full text