Fast quantitative parameter maps without fitting: Integration yields accurate mono-exponential signal decay rates

Autor: Ruitian Song, Joseph L. Holtrop, Claudia M. Hillenbrand, Jane S. Hankins, M. Beth McCarville, Ralf B. Loeffler
Rok vydání: 2017
Předmět:
Zdroj: Magnetic Resonance in Medicine. 79:2978-2985
ISSN: 0740-3194
DOI: 10.1002/mrm.26964
Popis: Purpose To develop a computationally fast and accurate algorithm for mono-exponential signal modelling and validate the new technique in the context of R2* mapping for iron overload assessment. Methods An algorithm is introduced that directly calculates R2* values from a series of images based on integration of the mono-exponential signal decay curve. The algorithm is fast, because fitting is avoided and only arithmetic computations without iterations are applied. Precision and accuracy of the method is determined in comparison to the conventional log-linear (LL), nonlinear least-squares-based Levenberg–Marquardt (NLM), and squared nonlinear Levenberg–Marquardt (SQNLM) methods, which rely on iterative curve fitting. Results In simulations, the signal integration based method consistently had the same or better accuracy than the LL, NLM, and SQNLM algorithms for R2* values ranging from 50 s−1 to 1200 s−1. In phantoms and in vivo (12 participants), this method was robust over a wide range of R2* values and signal-to-noise ratios. Computation times were approximately 100, 1460, and 930 times faster than those of the LL, NLM, and SQNLM methods, respectively. Conclusions The fast signal integration method accurately calculates R2* maps. It has the potential to replace conventional, mono-exponential fitting methods for quantitative MRI such as R2* parameter mapping. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.
Databáze: OpenAIRE
Externí odkaz: