Neonatal 3T PRESS Basis Sets (MARBLE)

Neonatal 3T basis sets used for the multi-site MARBLE study, simulated using Vespa. These correspond to the vendor-provided PRESS sequences used on Siemens, Philips and GE scanners, as described in Lally PJ, Montaldo P et al. Lancet Neuro 2019.

For each vendor, a basis set is provided for each of TE=60ms and TE=288ms.

File types:

• .RAW files (compressed into RAW.zip) - includes additional metabolites which were not included in the final basis set, but could still be useful for other neonatal MRS studies.
  *In GSH, glycine component has an additional 10Hz Gaussian broadening applied, following Kaiser LG. et al. JMR 202.2 (2010): 259-266*
• .in files - makebasis input file used to generate the fitting basis set from .RAW files
• .basis files - fitting basis set.
  *For MARBLE, the following metabolites were omitted from the fit with 'CHOMIT' control parameters: AcAc, Acn, Etm, Ser. The following metabolites were also combined with 'CHCOMB' due to their strong correlation during fitting: Thr+Lac, Asc+GSH*
• .pdf files - visualisation of the basis set components

Some specific details:

• Narrow FWHM (1Hz Gaussian broadening) to allow accurate quantification of sharp linewidth neonatal spectra and undoped MRS phantoms.
• Two versions of each basis set: i) each metabolite has a single component in the basis set (as usual); ii) singlet peaks of NAA, Cho, Cr are separated from other resonances to allow two-point T2 relaxometry between TEs (as used for MARBLE)
• Ideal RF pulses are assumed
• TE1,TE2 are adjusted according to vendor/TE
• Reference singlet at 0ppm

If using this to calculate peak area ratios at TE=288ms, multiply the LCModel provided 'concentrations' by a factor of the number of protons that contribute to each peak (i.e. 3 for NAA singlet @ 2.0ppm, 3 for Cr singlet @ 3.0ppm, 3 for Lac+Thr doublet @ 1.3ppm, 9 for Cho singlet @ 3.2ppm) - this will give you relative peak areas which are consistently scaled.

Contact: p.lally [at] imperial.ac.uk