Skip to content

SegDict Utility (GUI)

Web App

The SegDict Utility web app provides a way to view and evaluate segments saved in the segDict that result after running Motion Correction and Segmentation.

Requirements

  • System service setup (via systemd)
  • A port available for web access

Key Features

Component Evaluation

  • View and evaluate individual components from the segDict
  • Accept/reject/undecided classification for each component
  • Multiple evaluation methods:
    • CNMF Component Evaluation
    • Accepted Labels from segDict (ISX)
    • Neural Network Component Evaluation
    • Previously Done Accepted/Rejected Components

Visualization Tools

  • Spatial Profile View:

    • Display component spatial footprints
    • Toggle cell number overlay
    • Show all components or only accepted ones
    • Multiple colormap options
    • Integration with downsampled images

  • Temporal Profile View:

    • Display calcium signals for each component
    • Peak detection with two algorithms:
      • Scipy-based peak detection
      • Iterative differences method
    • Signal processing options:
      • Baseline adjustment
      • Smoothing via Savitzky-Golay filter
      • Spike deconvolution visualization
    • Peak transient profile analysis

Neural Network Integration

  • Model Training:
    • Add sessions to training set
    • Train new models with optimized hyperparameters
    • Cross-validation performance metrics
  • Model Management:
    • Save trained models
    • Load existing models
    • Evaluate components using trained models

Export Capabilities

  • Component Selection Export:
    • Save accepted/rejected/undecided components as JSON
    • Preview export data
  • Peaks Dictionary Export:
    • Save peak detection parameters and results
    • Include baseline and smoothing settings
    • Preview peak data

Access

The web app can be accessed at http://[server-address]:[port]/segdict. Based on what is in Application settings, the default port is 8504.

Usage

  1. Session Selection
  2. Search and select a session containing segDict files

  3. View session details and component counts

  4. Component Evaluation

  5. Select components using the ROI selector
  6. Classify components as accepted/rejected/undecided

  7. Use existing evaluations or create new ones

  8. Visualization

  9. Toggle between spatial and temporal views
  10. Adjust visualization parameters

  11. Analyze peak characteristics

  12. Export

  13. Export component selections
  14. Export peak detection results
  15. Preview export data before saving