Brightness Shifts in Miniscope Videos + CellReg Issue

[Sarvesh30]

Hey everyone,

I’m working on cell registration across multiple sessions using CellReg (https://github.com/zivlab/CellReg) with data from UCLA miniscopes, and I’ve run into two related issues:

Brightness shifts within a session
On a few occasions, our recordings were unintentionally split into two (or more) AVI files—either because the cable got tangled and we had to pause and restart, or because the acquisition stopped unexpectedly. When we simply concatenate these segments, the baseline fluorescence and apparent activity levels differ noticeably between them, which complicates downstream analyses and risks biasing our results.

CellReg registration failures
These brightness differences seem to confound CellReg’s matching algorithm, causing it to fail or produce inconsistent alignments across sessions.

Has anyone encountered similar brightness-drift or video-segmentation issues? What strategies have you found effective for:

Normalizing or harmonizing brightness across split videos before registration?

Incorporating brightness correction into your CellReg pipeline or downstream analysis?

Any advice—whether at the preprocessing stage (e.g., histogram matching, flat-field correction) or during/after registration—would be greatly appreciated!

Thanks in advance for your insights.

Here is a comparison in brightness between video 1 and video 2 for reference: