How can I map my entire mouse brain MERFISH section to the Allen Brain Atlas and annotate the brain regions?

[kulansam]

Thank you for a amazing tool.

I have followed the tutorial at https://jef.works/STalign/notebooks/merfish-allen3Datlas-alignment.html, to annotate the brain regions in my own sample. But I am unsure which slice number I should use for whole-brain annotation. When I use slice number #177, the annotation looks different from what is shown on the website.
Basically, I want to annotate each single cell into respective brain regions.

Could you please help me with this?

[mmganant]

Hello,
Can you please share the image of the slice you are trying to align?
Best, Manjari

[kulansam]

Hello, Can you please share the image of the slice you are trying to align? Best, Manjari

[kulansam]

@mmganant, could you please help with this?

I have been using different slices for annotation, but each time, I seem to lose some regions. What would be the most robust and comprehensive approach to ensure that I capture all the regions in the brain?

I would greatly appreciate your help. Thank you!

[mmganant]

Hello,
It might be helpful if you could also provide an image of what you transform looks like/what regions you are missing as my answer may change.
I would not use slice #177 as an initial approximation, as it is much more posterior than the slice you are using. I would choose a section that more closely matches you image. In terms of parameters, in general, you could try decreasing the sigmaM parameter for the images to match up better. You could also increase the number of iterations you run the transform for.
It also seem like you have a couple holes in your slice and a big one in the hypothalamic area -- since STalign is a smooth transform, it can't deal with holes/tears within the tissue, so those regions may not align well.
Hopefully this helps! Manjari

[kulansam]

Thank you for your response. I will play with sigmaM parameter with lower value.

I would like to map my MERFISH slice against the following Allen atlas slice (#750, from ccf_2017-ara_nissl_10.nrrd):

After mapping, I found that only 174 regions were annotated in my MERFISH slice, while the Allen atlas slice included annotations for 1,328 regions (from allen_ontology.csv). I found more than 85% of regions were NOT annotated in my slice. The missing regions in my data include the CTX, isocortex, MPO, DCN, PFL, and others. Could you please help to resolve this problem.

During mapping, I have used the following parameters:
sigmaA = 2 #standard deviation of artifact intensities
sigmaB = 2 #standard deviation of background intensities
sigmaM = 2 #standard deviation of matching tissue intenities
muA = torch.tensor([3,3,3],device='cpu') #average of artifact intensities
muB = torch.tensor([0,0,0],device='cpu') #average of background intensities

The model output is :

Thank you,
Kulandai

[mmganant]

Hello, from the diagram it seems like you are not seeing brain regions you expect because there is still misalignment. I would suggest to investigate the intensity values of your own image you are trying to align and assigning sigmaA, M and muA and B parameters appropriately, as this is affects your transform significantly. You can find the description for these parameters on here: https://jef.works/STalign/STalign.html#STalign.STalign.LDDMM