How does motion affect GABA-measurements? Order statistic filtering compared to conventional analysis of MEGA-PRESS
Sofie Tapper1,2, Anders Tisell1,2,3, Peter Lundberg1,2,3
1Center for Medical Image Science and Visualization, Linköping University, Linköping, Sweden.
Purpose:
The aim of this study was to evaluate two post-processing techniques applied to MRS MEGA-PRESS data influenced by motion-induced artifacts. In contrast to the conventional averaging technique, order statistic filtering (OSF) is a known method for artifact reduction. Therefore, this method may be suitable to incorporate in the GABA quantification.
Methods:
Twelve healthy volunteers were scanned three times using a 3 T MR system. One measurement protocol consisted of two MEGA-PRESS measurements, one reference measurement and one measurement including head motions. The resulting datasets were analyzed with the standard averaging technique and with the OSF-technique in two schemes; filtering phase cycles 'RAW PC' and filtering dynamics 'RAW Dyn'.
Results:
The datasets containing artifacts resulted in an underestimation of the concentrations. There was a trend for the OSF-technique to compensate for this reduction when quantifying SNR-intense signals. However, there was no indication that OSF improved the estimated GABA concentrations. Moreover, when only considering the reference measurements, the OSF technique was equally as effective as averaging, which suggests that the techniques are interchangeable.
Conclusion:
OSF performed equally well as the conventional averaging technique for low-SNR signals. For high-SNR signals, OSF performed better and thus could be considered for routine usage.
