Forum Replies Created
-
Posts
-
I made a series of basis sets for fatty acids for PRESS at vartious echo times. Now I managed to build a set or metabolite list with six fatty acid spectra at TE 20ms. To do this for all TE I would have to go through the same tedious procedure again. I wish I could just keep the names, number of spins, damping and concentration rations and attach a new simulated spectrum for each of the metabolites, but now at a different echo time (or perhaps a simulated STEAM spectrum).
I tried to decipher the metabolite list files (*.ml) but I don’t see how the spectral data are inserted in these files (but it appears that they are).
This is the first 50 doubles fo the *.ml files I made
” ‘¬í z alpha_linolenic_18_3_Om3 sig#0′”
The rest is gibberish, so I suspect that we get a different number format from there on. perhaps the data? Can anyone tell me what the format of metabolite files is? Or, even better, point me to Matlab or Python code that reads and writes *.ml files for QUESTRonald Ouwerkerk
National Institutes of Health (NIH)Hi Minh,Anindita,
Sorry for the late reply. NIH was throwing jmrui e-mails in the spam-quarantine.
The prior knowledge and starting value files cannot be edited outside jMRUI. There is a new text export feature that lets you export them to a text file which easily reads into Excel as comma-separated values.
I have been trying to read the .pk and .sv files in Matlab and in a binary read you will recognize the peak labels, but I cannot decipher the file format enough to back-engineer the file format. Would be great if we got the Java or C++ code made public.Regards,
RonaldRonald Ouwerkerk
National Institutes of Health (NIH)Hi Joshua,
You guessed it! The signal amplitude is the amplitude in the time domain and this is equal to the frequency domain peak area. If you want the peak amplitude, just divide the area (=time domain amplitude) by the width (Area = height * width at half maximum). The result is related to the unit of peak width (Hz), not an integral of the intensities over the peak foot-to-foot, but peak absolute peak areas are meaningless anyway.This method yields peak heights relative to the baseline (= the fit residual)
Bye
Ronald Ouwerkerk
Ronald Ouwerkerk
National Institutes of Health (NIH)Hi,
I have given up on batch processing a long time ago. I do however process my data in batches. Similar spectra are just grouped in directories. If I varied a parameter in this series (e.g. TE) I have this information in the filename. It would make more sense to have it in the metadata or header, but as far as I know, that information is not available in any of the data export formats formats of jMUI.
I read the whole set and then phase correct all or individual (as needed), correct frequency offsets etc..
I found that batch processing does not work for this step because the same parameters are used for each step.
After I save the corrected spectra (Save in separated files) I do an AMARES fit.
I save the results in HTML (publish-all results) format because the HTML file has the individual filenames of the spectra. The text export does not (I wish the MRUI format header data: patient name, date, the frequency correction and the reference PPM and Hz were also part of the output).
I then use a Matlab script to read the html code and collect the data. Data are then saved in a comma separated value file with one spectrum per row and columns for amplitude frequency etc for each resonance fitted and a column for noise estimate. I wish there was a more direct way to obtain this output format because especially in v6.0beta the HTML produces an absurd amount of graphics files.
For batch processing to really work one would have to be able to do the phase correction/frequency shift automatically in a reliable fashion. Even though AMARES allows some shifts in frequency and fits the phase I find that results are better when all spectra are properly aligned and frequency shifts are only reliable on properly phased spectra.Ronald Ouwerkerk
National Institutes of Health (NIH)
