Comparing analysis methods in assessing dynamic dual bolus cardiac magnetic resonance perfusion flow
© Muller-Bierl et al; licensee BioMed Central Ltd. 2013
Published: 30 January 2013
We compared the analysis methods by correlation analysis, by comparing the medians of the measured flow values, by regression analysis, by Bland-Altman analysis, and by related samples Wilcoxon rank test. We also investigated numerically the throughput of noise in the TRF on the analysis methods 1-4.
Five methods (Fermi, Deconvolution, the Tofts methods, and Uptake) were found to be suitable for normal perfusion flow evaluation. Repeatability of the results - independent on the analysis methods - was bad, which is due to the movement of the septum. For measuring stress perfusion, numerical modeling shows that using the Fermi and modified Tofts methods results in large bias errors in the presence of noise in the TRF.
As long as there is noise present in the TRF due to the moving septum, the evaluation of perfusion flow is not possible. We expect that measurement of cardiac perfusion flow will be possible using the Turbo FLASH sequence and the theoretical frameworks for perfusion flow analysis using the dual bolus method. Based on our investigation so far, we claim that Fermi, Model-Free Deconvolution, Uptake and the Tofts methods are all suitable for normal perfusion flow evaluation. Evaluation of the TRF in heart has to use spatial registration and correction: To establish a reliable procedure to assess the perfusion value based on the dual bolus data, motion correction of the septum by spatial image reconstruction is mandatory and should be the next logical step.
The work was payed by the Department of Radiology of the Flemish University Hospital Brussels.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.