Registering partial 2D IHC images to Allen Brain Atlas using QuickNII

Science
1

Hi all,

I’m wondering if anyone else has successfully used QuickNII to register IHC images of specific brain regions to the Allen Brain Atlas. We have an IHC dataset of just the hippocampal formation, with 3-4 channels each.

I’m having a problem where the resulting overlay is too pixelated and less precise (see below). Likely due to it being optimized for whole brain slices.

CTRL6_3_ss_s001_overlay
CTRL6_3_ss_s001_overlay1920×1291 335 KB

We have about 20 brains in our data set, a Zeiss LSM800, 4 channels of interest, and about 3 hours/week to take images, so and it just takes way too long for us to get whole brain images.

Any suggestions/tips/thoughts?

Much appreciated!

m.

2

An expert on histology at the Allen Institute had the suggestions below. Feel free to post again if this doesn’t address your question, and I’d also encourage others from the community to comment.

“Maybe they can try the following approaches:

  • Instead of aligning the cropped hippocampal image directly in QuickNII, they should load a larger field of view that includes nearby anatomical landmarks (even if these are not stained in their IHC channels).

  • Use a low-resolution reference scan (e.g., brightfield or autofluorescence) of the entire section for the initial QuickNII alignment, then crop and overlay the high-resolution hippocampal channel data afterward. This keeps the coordinate mapping accurate while preserving the full resolution of the microscopy images.

  • Ensure the images are saved in TIFF format with minimal compression before import. Avoid resizing prior to loading into QuickNII—let QuickNII handle scaling. If possible, increase the “texture resolution” setting in QuickNII’s preferences to reduce atlas texture pixelation (note: this requires more GPU memory).

  • Use QuickNII for the rough alignment to the Allen Brain Atlas, then export the transformation parameters to VisuAlign, which allows finer local warping and scaling—especially useful for partial-brain datasets.

Let me know if you need anything else.”

3

Thanks, Jeremy and histology expert. I’ll try these options and report back. (I didn’t realize that QuickNII could take TIFF files! I thought they were only .png or *.jpeg. So trying this one first!)