Noise evaluation in transmission electron microscopy (TEM)

Scrips for evalutating the important noise parameters and correlations in TEM. Reconstruction of the point spread function from signal noise

The files within this folder relate to and are part of the supplementary material of

Zietlow, C., Lindner, J.K.N.

An applied noise model for scintillation-based CCD detectors in transmission electron microscopy.

Sci Rep 15, 3815 (2025).

https://doi.org/10.1038/s41598-025-85982-4

So, if you use any of these scripts for your evaluations please cite the above paper.

Please note, however, that the authors disclaim any liability for damages or consequences arising from the use of the provided scripts.

Users are responsible for ensuring the proper functioning and application of the code.

Requires the package:

-Yun Pu (2025). Waitbar for Parfor (https://www.mathworks.com/matlabcentral/fileexchange/71083-waitbar-for-parfor)

- Andreas Korinek (2025). Gatan Digital Micrograph file reader (https://www.mathworks.com/matlabcentral/fileexchange/45933-gatan-digital-micrograph-file-reader)

Requires file format:

-".dm4" from Gatan

"_Create_" files provide .mat files that are necessary to load into other scripts! This is explained in the script at hand.

"_Analyse_" files just provide figures.

The input parameters that are to be selected by the user are briefly explained in the beginning of the respective scripts. The parameters are found thereafter. Everything else is automated.

The files are sorted by the number of the figure they are first used for and relate to the above paper . Therefore, the scripts work in the order that they are provided in.

The file "Create_Brighter_Fatter_Binning" is only for evaluation of the binned signal noises. It does not appear as a figure in the manuscript.

Since the Pearson coefficients of the signal are determined at the target intensity of the gain reference, this requires twice the number of frames as were used for creating the gain reference.

Therefore the number of signal and for background frames in the data folders for the gain ref should be at least twice the number of frames actually used for the gain reference.

As an example: for our evaluation, we had acquired 66 signal and 66 background frames at the intensiy of the gain reference, out of which 30 signal and 30 background frames were used to create the actual gain reference.

Therefore 60 Signal and 60 Background frames were used for the evaluation of the Pearson correlation coefficients of the signal at the intensity of the gain reference.

Here, 30 signal and 30 background frames were used for the evaluation of the k-value Pearson coefficients and 30 signal and 30 background frames were used for the Pearson coefficients of the signal itself.

Thus, the evaluation requires two folders, one for the background files and one for the signal files, with at least 60 files each. The 6 excess files, each, are required to correct for cosmic rays.

For the bias frame, 55 non-subtracted dark frame files at zero-exposure were acquired, out of which 30 were actually used for the creation of the bias frame.

For the analysis of the detector noises, we acquired 49 background-subtracted dark frame files with increasing acquisition times.

For the regression analysis of the signal noises, 160 non-subtracted dark frames and 160 non-subtracted signal frames of increasing acquisition time were acquired.

For the regression analysis of the signal noises under binning, a reduced set of 113 non-subtracted dark frames and 113 non-subtracted signal frames of increasing acquisition time were used to mitigate non-linerity effects.