NoRMCorre: An online algorithm for piecewise rigid motion correction of calcium imaging data

Area of Science:

• Neuroscience
• Computational Biology
• Image Analysis

Background:

• Motion correction is critical for analyzing calcium imaging data.
• Non-rigid motion artifacts in two-photon microscopy hinder accurate data interpretation.
• Existing methods struggle with computational performance and large fields of view.

Purpose of the Study:

• To introduce a fast and efficient algorithm for non-rigid motion correction in calcium imaging.
• To address limitations of existing motion correction techniques.
• To provide a scalable solution for large-scale image registration problems.

Main Methods:

• Developed Non-Rigid Motion Correction (NoRMCorre) algorithm based on template matching.
• Employed spatial patch splitting and sub-pixel registration for accurate alignment.
• Utilized up-sampling of alignments to generate smooth motion fields for piecewise-rigid approximation.

Main Results:

• NoRMCorre achieves real-time performance, comparable to or faster than existing methods.
• The algorithm effectively corrects non-rigid motion artifacts in simulated and in vivo datasets.
• Open-source Matlab and Python code are available for broader research application.

Conclusions:

• NoRMCorre offers a robust solution for non-rigid motion correction in calcium imaging.
• The method is particularly beneficial for large-scale image registration tasks with complex deformations.
• The provided open-source code facilitates its adoption in the neuroscience community.