A KNOWLEDGE-BASED APPROACH TO MINIMIZE BASE-LINE ROLL IN CHEMICAL-SHIFT IMAGING

A method has been developed to minimize baseline roll in chemical shift imaging (CSI). The technique is fully automated and employs knowledge based data processing in the frequency domain. The key feature of the algorithm is the computation of the ''trough'' and ''ripple'' components in the CSI data. The baseline roll can be regarded as an artifact that appears as a result of the summation of several sinc functions. Using prior knowledge, a mirror component corresponding to the artifact is created and added to the delayed spectrum. The method compensates for noise and zero-order phase error when computing the roll artifact. The results obtained on implementing the baseline roll minimization procedure on simulated time-delayed spectra indicated that the peak heights and areas were between 91% and 97% in magnitude when compared with the same peaks in the nondelayed spectra. The correction procedure was also assessed on clinical in vivo spectra. Nonlocalized P-31 MR spectra of the liver were obtained with and without an acquisition delay of 2.1 ms, and the time delayed spectra subjected to the baseline minimization routine. Metabolite peak heights and areas in the corrected spectra were approximately 94% in magnitude when compared with the same peaks in the original nondelayed whole volume spectra. Implementation of the baseline minimization procedure on in vivo localized spectra with varying signal to noise ratios produced good results. It takes approximately 13 s to implement the baseline roll minimization procedure. In this paper, the technique will be referred to as BaseLine Artifact Suppression Technique BLAS routine.