Fast and accessible morphology-free functional fluorescence imaging analysis.

Area of Science:

• Neuroscience
• Computational Biology
• Data Science

Background:

• Optical calcium imaging records neural activity at various scales.
• Traditional analysis methods rely on spatial features, limiting generalizability.
• Graph Filtered Temporal Dictionary Learning (GraFT) uses temporal activity for component identification.

Purpose of the Study:

• To present advancements to the GraFT algorithm for improved efficiency and scalability.
• To enhance the accessibility of GraFT through a graphical user interface.
• To demonstrate GraFT's utility across diverse imaging modalities.

Main Methods:

• Integrated an efficient L1 LASSO solver for faster computation.
• Applied compressive sensing and random projections to reduce data dimensionality and complexity.
• Developed a graphical user interface for user-friendly operation and analysis.

Main Results:

• Achieved substantial speedups in GraFT algorithm execution while maintaining analytical accuracy.
• Demonstrated increased scalability for larger and more complex calcium imaging datasets.
• Validated GraFT's applicability to meso-scale, vascular, and axonal imaging data.

Conclusions:

• Advanced GraFT offers a more generalizable and computationally efficient approach to analyzing neural activity from calcium imaging.
• The enhanced algorithm and user interface increase accessibility for researchers across various neuroscience domains.
• GraFT proves effective for analyzing diverse neural and vascular dynamics beyond traditional somata-centric methods.