Noise in Tiff Image Processing (Non-zero Values in Empty Regions Affecting Marker Expression Analysis)

[vyshakha]

I'm processing a Tiff image with 19 channels, each representing different markers in the cytoplasm and nucleus. However, I’ve noticed that every pixel in each channel has a non-zero value, even in regions where there is no tissue (empty areas or background). This results in significant noise in the data.
The next step involves analyzing the marker expression for each channel in every cell of the tissue. However, due to the presence of this noise, the expression values appear to be very high, and even markers that should not be present in certain cells are showing values.
Is there a way to reduce or eliminate this noise, especially in non-tissue regions, to ensure more accurate marker expression analysis? Any suggestions or solutions would be greatly appreciated.

[alex-l-kong]

Hi @vyshakha did you use the toffy you can try using our toffy repo (https://github.com/angelolab/toffy), specifically the Rosetta components, to denoise data. If you used a different pipeline to extract the data, however, you may want to check and see that the data has been denoised correctly.

Are a lot of non-zero values very close to zero, or are the non-zero values significantly high enough?