tSNR gain maps vs % variance explained

[Tasha-P]

Dear Lars,

In the Neuroscience methods paper, the tSNR gain maps are expressed in %. I found this a little confusing since the provided explanation for tSNR gains was a reduction factor in the standard deviation, which is "closely" related to the relative explained variance. How then do you interpret the outputs from tapas_physio_compute_tsnr_gains? e.g., if I have a value of 1.2 does that simply mean I have a 20% increase in tSNR?

Thank you,

T

[mrikasper]

Dear @Tasha-P,

I believe you are right, i.e., that the tSNR gains are just relative to the baseline tSNR without using the regressors. Thus, a value of 1.2 would be a gain of 20%.

If you look at the code of tapas_physio_compute_tsnr_gains, you will see in the header it labels the outputs as [tSnrImageArray, fileTsnrArray, ... tSnrRatioImageArray, fileTsnrRatioArray], which means one of the output files will be the absolute tSNR, and the other one relative gain (or ratio) compared to an uncorrected time series.

I am sorry that this is not clearer in the documentation and hope this explanation helps.

All the best, Lars