Main features

jMRUI is a software package for the processing and analysis of magnetic resonance spectroscopy datasets, either one-dimensional spectra recorded in clinical scanners or in high-field spectrometers (e.g. HR-MAS NMR spectra), or spectroscopic imaging (MRSI) datasets.

jMRUI offers:

  • Black box quantitation algorithms based on singular-value decomposition (SVD): The state space methods HSVD, HLSVD, and HTLS and the linear prediction method LPSVD. These non-interactive black box techniques are efficient for quantitating signals with good signal-to-noise ratios. They are also helpful in parametrizing signals of unknown composition and shape, but they cannot make use of all available prior knowledge.
  • Non-linear least-squares (NLLS) quantitation algorithms: AMARES, QUEST and AQSES. AMARES is an improved version of VARPRO enabling to impose prior knowledge on the model-function parameters. QUEST and AQSES are based on the availability of a metabolite signal basis set. They differ in the way how they treat the signal background.
  • Pre-processing algorithms such as rapid removal of dominant signals using HLSVD and HLSVD-Pro, or time–frequency analysis, the Cadzow enhancement procedure for noise reduction, the ER-filter for frequency selection, and Gabor tools for peak extraction and dynamic phase correction etc.
  • Simulation of metabolite basis sets: quantum-mechanical signal simulators NMR-SCOPE and NMRScopeB based on numerical solution of Liouville-von Neumann equation (by BCH expansions or eigenvalue calculations, respectively). NMRScopeB serves also as a tool for pulse sequence design. Signal simulation from a model function is also available.
  • Optimized quantitative MRS for clinical routine: DICOM data transfer,  absolute quantitation, support for normal value database.
  • Spectra visualisation: single spectrum, series of spectra, CSI with an MRI image.
  • Conversion routines for data files from most manufacturers: Bruker, General Electric, Philips, Siemens, Varian, etc. Moreover, jMRUI handles the new advanced DICOM format for MRS, MRSI and MRI.


In jMRUI parameter estimation of MR Spectroscopy data is carried out in the time domain.

Parameter estimation of MR Spectroscopy data can be carried out either in the time domain or in the frequency domain,

  • Time domain methods carry out operations directly on the measured FID or spin-echo signal.
  • Frequency domain methods process the Fourier spectrum of the original signal.


Distortions of the FT spectrum due to imperfections in the measured time domain signal can cause problems for frequency domain analysis methods. This specially holds true for in vivo MR signals:

  • data lengths are short due to relatively fast relaxation decay and limited acquisition time, giving rise to truncation artifacts in the discrete FT spectrum
  • missing initial data points (e.g. phase encoding sequences) necessitate back-extrapolation or sinc-deconvolution methods to produce a phased spectrum with correct baseline
  • quickly decaying signals (i.e. short relaxation times) from immobile compounds or instrument switching can give rise to broad background features in the FT spectrum, which underlie the resonances of interest, and lead to an overestimation of those areas, if careful precautions are not taken. Baseline correction methods exist, but introduce more approximations, possible operator dependencies and require extra processing time.

These problems can be handled with less approximations and assumptions using time domain analysis, the biggest advantage being that the raw data is processed without any preprocessing steps, such as windowing, apodisation, deconvolution, baseline correction and phasing.


  • ER Filter – Cavassila, S., Fenet, B., van den Boogaart, A., Rémy, C., Briguet, C., Graveron-Demilly, D. ER-Filter: a preprocessing technique to improve the performance of SVD-based quantitation methods. Journal of Magnetic Resonance Analysis 3: 87-92, 1997.
  • HLSVD water filtering – van den Boogaart, A., van Ormondt, D., Pijnappel, W.W.F., de Beer, R., and Ala-Korpela, M. (1994). In Mathematics in Signal Processing III, (J.G. McWhirter, Ed.), pp. 175-195, Clarendon Press, Oxford.