Meeting Banner
Abstract #1757

Downsampling Strategies for MRS Data

Alexander Fuchs1, Anke Henning, Peter Boesiger

1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland

MR acquired data is usually sampled at higher much higher frequencies than the Nyquist condition. Before further processing this data needs to be downsampled to reduce the burden on storage requirements and processing time. Two commonly used downsample strategies are applied to ideal spectroscopic data and different artifacts related to these methods are identified. Especially in spectroscopy, artifacts in the initial time points of FIDs or phase distortions can lead to problems in further quantitative evaluations. Therefore a method is proposed that avoids the described problems and gives more reliable downsampled data.

Keywords

absorb accurate amplitude analog anti appears applied arbitrary artifact artifacts averaging avoid avoids band bandwidth beginning blue brain chosen coil combination comm common condition consisted containing correction corresponds covering cutting damping degrees delay demand distortions domain done duration dynamic eddy either engineering evaluation even evolution experience expo extension fast filter filtered filtering filters fitting form free frequency furthermore gets half hence henning human ideal identified impair improved in vivo induced initial introduced involves last lead leads like linear magnitude mainly metabolites might model moving much nature needs negatively noise novel origin originate oversample oversampling pact peak peaks periodic peter prediction prevents prob problem problems processed processing quantification quantitative real recovered reduce reduced related relax reliability respect response sample sampled samples simulate simulated smooth spectra spectral spectrum squared starting step steps stern strategies subsequently suffer system systems target third typical unwanted usually water wiggle wiggles zero