Application of Self-Organising Maps in Automated Chemical Shift Correction of In Vivo H MR Spectra

  • Authors:

  • Juhani Pulkkinen;Mika Lappalainen;Anna-Maija Häkkinen;Nina Lundbom;Risto A. Kauppinen;Yrjö Hiltunen

  • Affiliations:

  • -;-;-;-;-;-

  • Venue:
  • IDEAL '02 Proceedings of the Third International Conference on Intelligent Data Engineering and Automated Learning
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Frequency shift differences in 1H MRSI spectra due to magnetic field inhomogeneities pose a problem, if automated lineshape fitting routines (LF) or artificial neural network (ANN) methods are used for spectral quantification. Use of self-organizing map (SOM) analysis for automated shift correction of long echo time (TE=270 ms) in vivo 1H NMR spectra of human brain is demonstrated. The map is obtained by training a SOM with proton spectra and the chemical shifts of the reference vectors were calibrated. The maps were then used for classification of spectroscopic imaging data and the calibration information for corrections of chemical shifts.