Computational investigation of fitting for calculation of signal dynamics from hyperpolarized xenon-129 Gas MRI

Abstract

Aim: Computational fitting methods were investigated to determine the most accurate fitting approach for the calculation of dynamic hyperpolarized MRI parameters.
 Materials and Methods: The signal decay of a time-series Hyperpolarized xenon gas MRI phantom was fitted to Bloch equations using three methods varying the fitting parameters for calculation of flip angle, α, and longitudinal relaxation time, T1. The first fitting method used an initial calculation of α before the fitting process. The second and third techniques used direct fitting of signal decay equations with and without upper-lower boundaries for calculation of α, and T1. Wilcoxon signed-rank test was used to investigate the statistical significance of the calculated parameters.
 Results: The first approach was the most accurate fitting technique that allowed direct calculation of α=8.65° in agreement to the third approach α=8.73±0.78°, 8.75±0.12°, 8.67±0.05°. Additionally, the standard deviation of the calculated T1 was lower than 1% (T1=103.2±0.04s) which was significantly more accurate than the second method (T1=90±30.2s and 135.7±10.3s) and the third method (T1=101.4±5.1s and 113.5±16.1s).
 Conclusion: The first technique provides repeatable and reliable calculation of signal decay parameters including α and T1 from the dynamic hyperpolarized gas MR images and more accurate than direct fitting methods.