Dear FSL experts
I have some question in calculating and understanding the temporal signal noise ratio bu using FSL scripts, here is the question;
From published papers and the suggestions from FSL JISCMail list: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;cc5ff611.1406 The usage used in calculating tSNR is as follows:
1) fslmaths preprocessed_data –Tmean preprocessed_data_mean
2) fslmaths preprocessed_data –Tstd preprocessed_data_std
3) fslmaths preprocessed_data_mean –div preprocessed_data_std snr_data
***: the preprocessed_data here should at least be motion corrected.
The 3) usage here means the dividing the mean signal over time by its standard deviation.
However, in combination with some papers which thresholded at 150 before calculating the tSNR and use 100 as lower threshold in identifying the acceptable SNR, I guess there are something I missed in tSNR calculation.
What is the reasonable value set as threshold in identifying the “bad” or “good” tSNR, if we use the scripts above ?
Besides, is it correct in calculating the tSNR by extracting the mean intensity of snr_data above, which is: fslstats snr_data –M or just use a thresholded mask as : fslstats snr_data –k mask_thresholded –M ?
For extracting the mean standard deviation, dose this “fslstats preprocessed_data_std –k mask –M ”
correct?
Any help will be of great appreciate.
Thanks
Qiuli
|