A validation framework for neuroimaging software: The case of population receptive fields
Fig 3
We analyzed BOLD data collected with an on-off visual contrast stimulus (8 cycles/run, four repetitions). Noise was assessed by comparing the BOLD time series with a harmonic at the same frequency. A) The probability density functions of the noise values of three voxels with low (green), mid (blue) and high (red) noise. B) The colored curves are the mean BOLD time-series plotted as percent modulations (mean of the four repetitions). The black curves are the eight cycle harmonic fit to the data. The noise is the difference between the two. The blue curve is the middle of the noise level found in visual cortex. The other two curves show low and high noise examples that are often found in real measurements. Low noise: SNR = 6.55dB; mid noise: SNR = -3.03dB; high noise: SNR = -6.68dB.