CaimanGUI

A graphical user interface for visualizing CaImAn processed imaging data.

CaImAn is a powerful computational software to process one-photon and two-photon imaging data. Here, CaimanGUI provides complementary visulizing functionalities that help neuroscientists to curate their processed imaging data.

Installation

conda create -n caiman caiman
conda activate caiman
pip install git+https://github.com/nctrl-lab/CaimanGUI

CaimanGUI is built upon pyqtgraph that is included in the CaImAn package. Thus, you can simply run CaimanGUI in caiman environment (assume you call caiman for your CaImAn installation):

conda activate caiman
caiman

Getting started

Using the GUI

CaimanGUI is mainly designed for visualizing CaImAn processed one-photon imaging data (should also work for two-photon data). The implemented functionalities are partly inspired by the widely used Suite2p software for two-photon data.

Neuron Table

The neuron table provides a Phy-like interface for viewing and managing neurons. It displays:

• ID: Neuron identifier
• Rval: Spatial correlation score
• SNR: Signal-to-noise ratio
• CNN: CNN prediction score (if available)
• Area: Number of pixels in spatial footprint
• Quality: Component quality based on CaImAn filtering (Good/Bad)
• Status: User-assigned status (Good/Noise/Uncertain/Unassigned)

The table supports:

• Sorting: Click column headers to sort (numeric sorting for ID, Rval, SNR, CNN, Area)
• Filtering: Use "View components" dropdown to filter by status (All, Accepted, Rejected, Uncertain, Unassigned)
• Selection: Click rows to select neurons (Ctrl/Shift for multiple selection)

Quality Filtering

The GUI includes quality thresholds for filtering components:

• Rval high/low: Spatial correlation thresholds
• SNR high/low: Signal-to-noise ratio thresholds
• CNN high/low: CNN prediction thresholds
• Area low: Minimum area threshold (in pixels)

Components must meet both low thresholds (AND) and at least one high threshold (OR) to pass quality filtering.

Keyboard Shortcuts

• Ctrl+R: Run CaImAn processing

• Ctrl+O: Open HDF5 file

• Ctrl+S: Save data

• Ctrl+Shift+S: Save as...

• Ctrl+W: Exit application

• ESC: Unselect all neurons

• Alt+G: Mark selected neuron(s) as Good (Accepted)

• Alt+N: Mark selected neuron(s) as Noise (Rejected)

• Alt+M: Mark selected neuron(s) as Maybe (Uncertain)

• Up/Down arrows: Navigate through neurons

View Modes

The GUI supports several viewing modes:

• reset: Initialization mode showing contours of the current group
• neurons: Show colormap of the current group, display multiple selected components
• correlation: Show single selected component and all other components colored by temporal correlation
• accepted: Display accepted components with keyboard navigation
• neighbors: Display neighbor correlation of accepted components

Motion-corrected movie

You can also visualize the motion-corrected movie (menu View/Movie) together with the corrected in-plane shifts and the fluorescence trace of a selected cell. Note that reading a F-order memory-mapped file is faster than a C-order memory-mapped file.

Outputs

CaimanGUI outputs include:

• accepted_list: List of accepted cell IDs (caiman_model.estimates.accepted_list)
• rejected_list: List of rejected cell IDs
• uncertain_list: List of uncertain cell IDs (needs further review)
• idx_components: Components passing quality filters (this is updated by the Quality Filtering criteria)
• idx_components_bad: Components failing quality filters

All data is saved in the HDF5 or NWB format, preserving the full CaImAn CNMF object structure.