Signal Drift and Calibration for Magnetic Resonance Spectroscopy

• Activity Number: 377
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 11:15 AM
• Sponsor: Mental Health Statistics Section
• Author(s): Laura Mariano* and Ben Rowland and John Irvine and Alexander Lin
• Companies: Draper Laboratory and Brigham and Women's Hospital and The Charles Stark Draper Laboratory and Brigham and Women's Hospital
• Keywords: biomarkers ; magnetic resonance spectroscopy ; sensor drift ; sensor calibration ; multimodal data ; machine learning

Abstract

Magnetic resonance spectroscopy (MRS) provides a non-invasive method for studying the levels of metabolites in the brain. To achieve acceptable signal levels, many identical experiments are performed and the results combined. There is expected to be some main frequency drift observed across the set of sequentially-acquired averages, but the effect of this drift is usually considered negligible and, therefore, not addressed by the vast majority of post-processing methods. However, recent analysis of the raw signals collected from each coil, prior to averaging and combination, revealed in some cases a significant frequency drift that had a major impact on the quality of the final post-processed signal. Failure to identify and correct for this frequency shift can seriously affect the suppression of the water peak and the extraction of signal peaks related to metabolites of interest. This presentation will describe the frequency drift, present a method for correcting the signal, and demonstrate the effects on the subsequent analysis of key metabolites.