Effect of FID-MRSI backward linear prediction with autoregressive algorithm on metabolite estimates for compensation of acquisition delay

Motivation: 1H Free-Induction Decay (FID) MRSI is limited by the acquisition delay (AD) between the RF excitation pulse and the FID signal. N initial data points are thus lost. Goal(s): Our goal was to evaluate the consistency of the Backward Linear Prediction (BLP) auto-regressive reconstruction method to recover the lost FID data points. Approach: In-vivo rat data were used to investigate the impact of the BLP methodology in a cut-and-recover approach; further Monte-Carlo simulations were used to identify the method validity limit. Results: In-vivo and Monte-Carlo results highlighted the consistency of the BLP methodology for realistic FID reconstruction ranges. Impact: Focusing on metabolites of interest, no significant variations of brain map concentrations have been detected between original FID acquisitions and BLP reconstruction outcomes between AD=1.3ms and AD=0.708ms. Moreover, Monte Carlo simulations showed good quantification reliability until AD=2.7 ms.