Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Denoising Two-Photon Calcium Imaging Data

Figure 3

Decomposition of fluorescence data into signal and noise components.

The representative time series data is from Cell 11, Pixel 45 in our data set. (a) Relative fluorescence data (blue) measured in three consecutive trials (dashed lines: trial boundaries). The fit (red) is the estimate obtained by the signal plus correlated noise model, containing both stimulus-evoked activity and noise. (b) Relative fluorescence data (blue) in three consecutive trials and estimate (red) of the signal component (i.e., the stimulus-evoked activity). (c) Autocorrelation function (red) of residual noise, , lies within the 95% whiteness bounds (blue). (d) The quantile-quantile plot of the residuals confirms Gaussianity. The results in (c) and (d) prove that the residuals are independently and identically distributed Gaussian, and the systematic variance in the data has been explained by the harmonic regression and autoregressive terms. (e) Orientation tuning curve obtained from the denoised signal estimate in (b). The SCN model provides a smooth fit to the across-trials mean of the data. Point-wise approximate 95% confidence intervals are also shown. The SCN model preserves the complex, asymmetric shape of the response tuning curve.

Figure 3

doi: https://doi.org/10.1371/journal.pone.0020490.g003