An automated algorithm for combining multivoxel MRS data acquired with phased-array coils

Nimrod Maril, Robert E. Lenkinski

Research output: Contribution to journalArticlepeer-review

26 Scopus citations


Purpose: To develop a fully automated algorithm for combining multivoxel magnetic resonance spectroscopy (MRS) data acquired with a phased-array coil. Materials and Methods: The frequency-domain fitting method of LCModel (Provencher SW, Magn Reson Med 1993;30:672-679) was utilized to analyze the individual data sets. The phase corrections and the metabolite areas were then extracted from the LCModel output files for each individual spectrum. These areas were used to determine the dominant metabolite for each spatial location and to combine the individual spectra in a weighted manner. Results: The combination of MRS data acquired from a phantom and the brains of normal volunteers with a four array coil yielded improved signal-to-noise ratio (SNR) in all voxels. The average improvement in SNR of the combined spectrum, as compared with the best of the individual spectra at each spatial location, was 1.4. In the phantom, the predicted SNR improvement of two-fold was achieved at the center of the sample. In the brain, the maximum improvement was 1.8, due to sampling of the ventricles in the center of the sample. Conclusion: The method described in this report provides a means for employing phased-array coils in MRS with the same advantages as those found in MRI.

Original languageEnglish (US)
Pages (from-to)317-322
Number of pages6
JournalJournal of Magnetic Resonance Imaging
Issue number3
StatePublished - Mar 2005


  • Automation
  • CSI
  • LCModel
  • Phased array coil
  • Weighted combination

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging


Dive into the research topics of 'An automated algorithm for combining multivoxel MRS data acquired with phased-array coils'. Together they form a unique fingerprint.

Cite this