Beta versions of Gannet on the way

Over the Christmas period, John and Nick have ironed out some remaining discrepancies between Cardiff code and the master Gannet version, including handling transatlantic differences such as default units for output figures.

We anticipate that this will lead to a beta version of Gannet that will be made available here by the end of January, which will be publicised to our current collaborators and freely downloadable.

Co-registering the MRS voxel to an image

This is 'the problem that will not die', and at various times, I have spent weeks on the problem without much to show for it. This document outlines the route for a solution.
As a taster, here is a triply rotate MRS voxel on a triply rotated T1 image. First the scanner screenshot:
 

And now the reconstructed voxel (overlaid on the image):

 

We may be nearly there.

Anonymizing Siemens .rda files

In working towards a useful distribution version with example data files, we need to make sure that all files are de-identified. This script does the trick (matlab):


a=dir('*.rda');
for number=1:size(a)
    a(number).name
    fid = fopen(a(number).name,'r+');
    tline = fgets(fid);
    position=ftell(fid);
    tline = fgets(fid);
    magic=tline;
    size(magic)
    for ii=14:(size(magic,2)-2)
      magic(ii)='*';
    end
    fseek(fid,position,'bof');
    fwrite(fid,magic,'char');
    fclose(fid);
end

Cross-vendor harmonisation: alpha and kappa

One aim we have to overcome differences in GABA concentration measurements across vendor.  Currently there are two main ways to quantify GABA signal:

1. As an integral ratio to Creatine in the 'off' spectrum.
2. As a concentration in institutional units (i.u.), relative to the unsuppressed water signal.

The main advantage of method 2 is that differences between scanners (mentioned below) can be accommodated to give values that can be compared across studies and across scanners.

The concentration c_G of GABA in i.u. can be expressed as:

where subscript G and w refer to water and GABA and where factors R account for T1 and T2 relaxation, kappa is the editing efficiency of GABA signal (typically around 50%) and alpha is the signal fraction of macromolecular signal.  In this framework, only two things change specifically by vendor* - the editing efficiency and the MM signal fraction and appropriate correction factors should result in cross-vendor comparable i.u. values.

The editing efficiency will depend on multiple factors including the slice-selective refocusing pulse spatial profile and bandwidth.  Finite bandwidth refocusing leads to a loss of edited signal efficiency related to the chemical-shift displacement (see e.g. our 2007 paper).

The MM signal fraction will depend on editing pulse selectivity. Although editing pulses are placed at 1.9 ppm in the GABA-editing experiment, they partially invert MM spins at 1.77 ppm, which are coupled to MM signals at 3 ppm, leading to a significant MM contribution to the edited signal.  The degree of this inversion at 1.7 ppm will depend on the inversion pulse shape and length, with length largely being limited by the ~70 ms echo time preferred for edited experiments and the B1.

So here is the plan - measure kappa for each MEGA-PRESS implementation and somehow get a handle on differences in alpha.  



* ..but we mustn't forget that the R terms vary by TE and TR which may vary site-to-site.

GannetLoad Processing Flowcharts

GannetFit Output

GannetLoad output