Skip to content

ICA Tab Guide

The ICA tab is the primary workspace for reviewing and classifying components. This guide explains each visualization panel and how to use them effectively.

Overview

The ICA tab presents multiple synchronized views of your ICA components. Selecting a component in any view highlights it across all visualizations.

Scatter Plots

Kappa vs Rho Plot

The main scatter plot shows components plotted by their two key metrics:

  • X-axis (Kappa): TE-dependent signal weight - higher values indicate more BOLD-like signal
  • Y-axis (Rho): TE-independent signal weight - higher values indicate more noise-like signal

Interpreting the plot:

Position Interpretation
High Kappa, Low Rho Likely BOLD signal (accept)
Low Kappa, High Rho Likely noise (reject)
Low Kappa, Low Rho Unclear - needs manual review
High Kappa, High Rho Mixed signal - needs manual review

Features:

  • Zoom: Scroll to zoom in/out
  • Pan: Click and drag to pan
  • Select: Click a point to select that component
  • Elbow lines: Dashed lines show tedana's automatic thresholds (if available)

Rank Plots

Two additional plots show:

  • Kappa Rank: Components ordered by kappa value
  • Rho Rank: Components ordered by rho value

Connecting lines between plots help visualize how a component's kappa and rho ranks compare.

Pie Chart

Shows the proportion of total variance explained by each classification category:

  • Accepted (green): Components kept in the denoised data
  • Rejected (red): Components removed as noise
  • Ignored (gray): Components set aside (if applicable)

The pie chart updates in real-time as you reclassify components.

Brain Viewer

The interactive brain viewer displays the spatial map of the selected component.

Display:

  • Mosaic view with 7 slices per orientation (axial, coronal, sagittal)
  • Statistical z-map overlaid on anatomical template
  • Color scale indicates component weight at each voxel

Controls:

  • Toggle between Interactive (Niivue) and Static (PNG) views
  • Interactive view allows slice navigation
  • Static view shows tedana-generated figures (if available)

Performance

If Rica feels slow, switch to static PNG view. Interactive Niivue rendering can be demanding on some systems.

Time Series

Displays the component's time course from the mixing matrix.

What to look for:

Pattern Interpretation
Smooth, oscillating Could be BOLD or physiological
Spiky, irregular Likely motion artifact
Slow drift Scanner drift or physiological
Step changes Motion or acquisition issues

The x-axis shows time (in TRs), and the y-axis shows the mixing weight.

FFT Spectrum

Shows the frequency content of the component time series.

What to look for:

Frequency Typical Source
< 0.01 Hz Scanner drift, very slow physiology
0.01-0.1 Hz BOLD signal range
0.15-0.4 Hz Respiratory (~0.3 Hz)
0.8-1.2 Hz Cardiac (~1 Hz)
High frequency spikes Motion artifacts

Note

Exact frequencies depend on your TR. The above assumes a typical TR of ~2 seconds.

Component Table

A sortable table showing all component metrics.

Columns include:

  • Component number
  • Classification (accepted/rejected/ignored)
  • Kappa and Rho values
  • Variance explained
  • Additional tedana metrics

Features:

  • Click column headers to sort
  • Click a row to select that component
  • Toggle table visibility with the collapse button
  • Option to keep original order or regroup by classification

Classification Controls

Toggle Switch

Three-state toggle for each component:

State Color Meaning
Accepted Green Keep in denoised output
Rejected Red Remove as noise
Ignored Gray Set aside (rare)

Keyboard Shortcuts

For efficient classification:

  • A - Accept current component
  • R - Reject current component
  • ← - Go to previous component
  • → - Go to next component

External Regressor Heatmap

If your tedana output includes external regressor correlations (tedana 24.1+), a heatmap shows how each component correlates with:

  • Motion parameters
  • Physiological regressors
  • Custom regressors you provided

High correlations with motion suggest the component captures motion artifacts.

Save Your Work

The Save button exports your classifications to a TSV file:

  • File name: manual_classification.tsv
  • Compatible with tedana's --manual-classification option
  • Contains original and manual classification columns

Save Often

Rica runs entirely in your browser. If you close the tab without saving, your classifications are lost!

Best Practices

  1. Start with accepted components: Review tedana's accepted components first to understand what "good" looks like in your data

  2. Use multiple views: Don't rely on a single visualization - check time series, FFT, and spatial maps together

  3. Look for patterns: Physiological noise often has characteristic spatial patterns (edge effects, ventricles)

  4. Trust but verify: Tedana's automatic classification is good, but not perfect - that's why you're reviewing!

  5. Document decisions: Consider keeping notes on why you accepted/rejected edge cases