Tracking

Tracking connects the “same” objects or particles across time-lapse sequences. Because objects may appear, disappear, divide, or change shape between frames, they typically receive different object IDs on each frame. Tracking solves this by assigning each object a persistent track ID that remains constant across frames for the same physical entity.

Overview of tracking methods

GA3 provides two distinct tracking approaches optimized for different applications:

Object tracking

Track Objects is designed for larger, morphologically complex objects like cells. It:

• Works directly with binary layers
• Uses object frame-to-frame overlap to assign track ID
• Handles objects that change shape between frames
• Combines easily with standard object measurements
• Is suitable when objects are morphing over time

Use for: Cells, organoids, cell clusters, tissue regions

Particle tracking

Track Particles is optimized for point-like particles and includes motion modeling:

• Works with centroid positions (not binary layers)
• Uses position-based algorithms with motion prediction
• Offers different motion models (Brownian, directed)
• Provides specialized motion analysis features
• Handles high particle densities efficiently
• Is suitable for small/rigid bodies

Use for: Vesicles, fluorescent spots, small organelles, diffusing molecules, single particles

This guide covers three complete tracking workflows demonstrating both approaches:

1. Track moving objects - Object tracking combined with morphology measurements
2. Particle tracking with motion analysis - Motion feature calculation and motility classification
3. Single particle tracking with diffusion - Mean squared displacement (MSD) and diffusion coefficient analysis

Common workflow structure

All tracking workflows share a similar pipeline structure:

Do not forget to enable tracking overlay

Workflow 1: Track moving objects

This workflow tracks cells or large objects while simultaneously measuring their morphological features.

  
---
config:
  look: handDrawn
  theme: neutral
---

graph TD
  subgraph "Tracking & Measurement"
    track[Track Objects]
    measure[Object Features]
    append[Append Columns]
    accum[Accumulate Tracks]
    chart[Line Chart]
    table[Table]
  end

  subgraph "Presentation"
    layout1[Stacked Layout - Left]
    layout2[Stacked Layout - Right]
    display[Display]
  end

  subgraph Outputs
    savechn[Save Channels]
    savebin[Save Binaries]
    savetab[Save Tables]
  end

  %% Main workflow
  channels[Channels] -->|Alexa 488| cellpose[Cellpose AI] -->|Cell| remove[Remove Touching Frame] -->|Cell| track
  remove -->|Cell| measure
  channels -.->|All| measure

  track -->|Records| append
  measure -->|Records| append

  append -->|Records| accum
  accum -->|Records| chart
  accum -->|Records| table

  %% Presentation
  chart --> layout2
  table --> layout1
  layout1 --> display
  layout2 --> display

  %% Outputs
  channels -->|All| savechn
  remove -->|Cell| savebin
  display -->|Results| savetab

Segmentation with Cellpose

Cellpose provides AI-powered cell segmentation on the fluorescent channel. The diameter parameter (35 pixels) is tuned for the typical cell size in the sample.

Border object removal

Remove Objects Touching Frame eliminates objects that touch the image border (5 pixel margin on all sides). This prevents tracking partial objects that enter or leave the field of view.

Object tracking

Track Objects assigns track IDs by matching objects between consecutive frames based on:

• Overlap between object positions
• Similarity in size and shape
• Proximity when objects don’t directly overlap

The output is a Records table with:

• TrackId: Persistent identifier for each tracked object
• Frame information: Time points where the object appears
• Object ID: Per-frame object number (changes each frame)

Object measurement

Object Features measures morphological and intensity features for each tracked object:

• Size: Area, perimeter
• Shape: Circularity, elongation, width, height
• Intensity: Mean, min, max, standard deviation in all channels
• Position: Centroid X, Y coordinates

Combining tracking with measurements

Append Columns merges the track IDs from Track Objects with the measured features from Object Features. Both tables have the same rows (one per object per frame), making the combination straightforward.

This creates a unified table where each object has:

• Its morphology and intensity measurements
• Its track ID linking it across time
• All temporal information

Track accumulation and filtering

Accumulate Tracks collects tracks across all frames with filtering:

• MinSegmentCount: 3 frames minimum
• Tracks appearing in fewer frames are discarded as transient noise

The accumulated data organizes objects by track, enabling temporal analysis of individual cells throughout the time-lapse.

Visualization

Line Chart displays trajectories color-coded by track, showing how each cell moves over time. The chart can plot any measured feature vs. time (area changes, intensity variations, etc.).

Table shows all tracks with measurements at each time point, allowing sorting by track ID or any measured parameter.

Results

This workflow produces:

• Track IDs linking objects across time
• Complete morphological measurements for each tracked object
• Trajectory visualizations
• Combined segmentation masks and measurement tables

Use cases: Cell migration tracking, cell growth analysis, morphology changes during differentiation, long-term single-cell analysis

Workflow 2: Particle tracking with motion analysis

This workflow focuses on motion characterization, classifying particles by their motility patterns and measuring detailed motion features.

  
---
config:
  look: handDrawn
  theme: neutral
---
graph TD

  subgraph "Tracking Pipeline"
    time[Time and Center]
    track[Track Particles]
    accum[Accumulate Tracks]
    motion[Motion Features]
    rolling[Rolling Average]
  end

  subgraph "Motion Analysis"
    trackfeat[Track Features]
    calc[Create Column]
    group[Group Records]
  end

  subgraph Objects
    chart[Line Chart]
    table1[Table]
  end

  subgraph Tracks
    scatter[Scatter Plot 4D]
    histo[Histogram]
    table2[Table - Tracks]
  end

  subgraph Presentation
    layout1[Stacked Layout - Left]
    layout2[Stacked Layout - Right]
    display[Display]
  end

  subgraph Outputs
    savechn[Save Channels]
    savebin[Save Binaries]
    savetab[Save Tables]
  end

  %% Main workflow
  channels[Channels] -->|Mono| spots[Bright Spots] -->|Bin| time
  channels -->|Mono| maxip[MaxIP]

  time -->|Records| track
  track -->|Records| accum
  accum -->|Records| motion

  %% Motion analysis branch
  motion -->|Records| rolling
  rolling -->|Records| table1

  accum -->|Records| trackfeat
  trackfeat -->|Tracks| calc
  calc -->|Tracks| group
  calc -->|Tracks| histo
  calc -->|Tracks| scatter
  group -->|Tracks| table2

  %% Presentation
  rolling -->|Records| chart
  table1 --> layout2
  table2 --> layout2
  chart --> layout1
  scatter --> layout1
  histo --> layout1

  layout1 --> display
  layout2 --> display

  %% Outputs
  channels -->|All| savechn
  spots -->|Bin| savebin
  display -->|Results| savetab

Spot detection

Bright Spots detects fluorescent particles with:

• Diameter: 9 microns (adjusted for particle size)
• Contrast: 18 (threshold for minimum brightness)

This produces a binary layer with detected spots in each frame.

Maximum intensity projection

MaxIP creates a composite image showing all particle positions throughout the time-lapse. This is an aid to show how the tracks align with actual positions.

Time and position measurement

Time and Center measures for each detected particle:

• Time: Timestamp of the frame
• CenterX, CenterY: Centroid coordinates

This table of positions becomes the input for particle tracking.

Particle tracking algorithm

Track Particles connects particles between frames using:

Parameters:

• maxSpeed: 1000 µm/frame - maximum distance a particle can move between frames
• maxGap: 2 frames - allows tracking through temporary particle disappearance
• motionModel: 1 (directed motion) - assumes particles have momentum

The algorithm predicts where each particle should appear in the next frame based on its previous motion, enabling accurate tracking even in crowded fields.

Track accumulation

Accumulate Tracks collects all tracks with:

• MinSegmentCount: 4 frames minimum
• Filters out short, unreliable tracks

Motion feature calculation

Motion Features calculates instantaneous motion parameters for each segment (position to position):

• Speed: Distance per time interval
• Direction: Heading angle
• Acceleration: Rate of speed change

These frame-to-frame measurements capture motion dynamics.

Rolling average smoothing

Rolling Average smooths noisy motion features over a window of frames. This reduces measurement noise while preserving real motion trends.

Track feature calculation

Track Features computes track-level summary statistics:

Per-track features:

• Segments: Number of frames in the track
• Length: Total path length traveled
• Duration: Time span of the track
• Speed: Mean, min, max speed
• Heading: Overall direction consistency
• Acceleration: Mean acceleration magnitude
• LineLength: Straight-line distance from start to end
• LineSpeed: Displacement divided by duration
• Straightness: LineLength / Length (1 = perfectly straight, 0 = highly curved)

These aggregate features characterize the overall behavior of each particle.

Motility classification

Create Column creates a calculated column classifying tracks by motility:

Classification logic:

Speed < 30 ? "STEADY" : (Straightness < 0.4 ? "MOTILE" : "PROGRESSIVE");

This produces three categories:

• STEADY: Low speed, minimal movement (confined or stationary)
• MOTILE: Higher speed but non-directional (random/diffusive motion)
• PROGRESSIVE: High speed and directional (directed migration)

The thresholds are adjusted based on the biological system.

Group by motility

Group Records aggregates tracks by their motility classification, producing summary statistics for each category (count, mean speed, mean straightness, etc.).

Visualization

Line Chart displays individual particle trajectories color-coded by track.

Scatter Plot 4D shows tracks as points in multidimensional feature space (speed vs. straightness, colored by motility class), revealing population structure.

Histogram shows the distribution of tracks across motility categories (how many steady vs. motile vs. progressive).

Tables display both per-frame motion data and per-track summaries.

Results

This workflow produces:

• Track IDs for all particles
• Instantaneous motion features (speed, direction, acceleration)
• Track summary features (length, duration, straightness)
• Motility classification
• Population statistics by motility class

Use cases: Vesicle transport analysis, cell migration modes, drug effects on motility, distinguishing active vs. passive transport

Workflow 3: Single particle tracking with diffusion analysis

This specialized workflow calculates mean squared displacement (MSD) to determine diffusion coefficients, essential for studying Brownian motion and molecular diffusion.

  
---
config:
  look: handDrawn
  theme: neutral
---
graph TD
  subgraph Tracking
    time[Time and Center]
    track[Track Particles]
    accum[Accumulate Tracks]
    table1[Table - Trajectories]
  end

  subgraph MSD
    msd[MSD]
    ungroup[Ungroup Records]
    reduce[Reduce Records]
    table2[Table - MSD per Δt]
    chart[Line Chart - MSD vs Δt]
  end

  subgraph Presentation
    layout1[Stacked Layout - Left]
    layout2[Stacked Layout - Center]
    layout3[Stacked Layout - Right]
    display[Display]
  end

  subgraph Outputs
    savechn[Save Channels]
    savebin[Save Binaries]
    savetab[Save Tables]
  end

  %% Main workflow
  channels[Channels] -->|Mono| spots[Bright Spots] -->|Mono| time
  channels -->|Mono| maxip[MaxIP]

  time -->|Records - Time, X, Y| track
  track -->|Records - TrackId| accum
  accum --> table1

  %% MSD analysis branch
  accum --> msd
  msd -->|Records| ungroup
  ungroup -->|Records| reduce
  reduce --> table2
  reduce --> chart

  %% Presentation
  table1 --> layout1
  table2 --> layout2
  chart --> layout3

  layout1 --> display
  layout2 --> display
  layout3 --> display

  %% Outputs
  channels -->|All| savechn
  spots -->|Mono| savebin
  display -->|Results| savetab

High-contrast spot detection

Bright Spots detects single fluorescent molecules or particles with stringent settings:

• Diameter: 0.69 microns
• Contrast: 500 (high threshold for bright, well-isolated spots)
• Intensity filters: Remove outliers

These strict criteria ensure only well-localized particles are tracked, critical for accurate diffusion analysis.

Tracking for diffusion analysis

Track Particles uses settings optimized for diffusive motion:

Parameters:

• maxSpeed: 4 µm/frame (tight constraint for slowly diffusing particles)
• maxGap: 2 frames
• motionModel: 0 (Brownian motion) - assumes random walk
• stdevFactor: 2.5 - statistical tolerance for position prediction

The Brownian motion model is essential for diffusion studies as it doesn’t assume directional bias.

Long track accumulation

Accumulate Tracks requires:

• MinSegmentCount: 20 frames minimum

Long tracks are necessary for reliable MSD curve fitting. Shorter tracks don’t provide enough data points for accurate diffusion coefficient calculation.

Mean squared displacement calculation

MSD computes the mean squared displacement for each track:

MSD formula:

MSD(Δt) = <[r(t + Δt) - r(t)]²>

Where:

• Δt: Time interval (lag time)
• r(t): Position at time t
• < >: Average over all time points and all tracks

The MSD is calculated for increasing time intervals:

• Δt = 1 frame: MSD between consecutive positions
• Δt = 2 frames: MSD between positions 2 frames apart
• Δt = 3 frames: MSD between positions 3 frames apart
• And so on…

Output table structure:

• Δt: Time interval
• MSD: Mean squared displacement at this interval
• StDev: Standard deviation of MSD
• Count: Number of measurements contributing to this point

MSD data processing

Ungroup Records separates the grouped MSD data structure for further analysis.

Reduce Records aggregates MSD values across all tracks, computing:

• Mean MSD: Average MSD at each time lag (the ensemble MSD)
• Count of tracks: Number of tracks contributing

This produces the final MSD curve for the entire population.

MSD curve visualization

Line Chart plots MSD vs. Δt, displaying:

• The characteristic MSD curve shape
• Error bars showing variability
• Linear or non-linear regime

Curve interpretation:

• Linear MSD(Δt) ∝ Δt: Normal diffusion (Brownian motion)
• MSD(Δt) ∝ Δt^α, α < 1: Subdiffusion (confined or hindered)
• MSD(Δt) ∝ Δt^α, α > 1: Superdiffusion (active or directed)

Diffusion coefficient calculation

For normal diffusion, the relationship is:

MSD(Δt) = 4D·Δt (2D motion)

Where D is the diffusion coefficient.

The slope of MSD vs. Δt gives: Slope = 4D

Therefore: D = Slope / 4

The diffusion coefficient units are typically µm²/s.

Fitting procedure:

1. Fit a linear regression to the initial linear region of the MSD curve (typically first 25% of time lags)
2. Extract the slope
3. Calculate D = Slope / 4

Results interpretation

Tables display:

• Per-track MSD curves
• Ensemble-averaged MSD
• Track counts at each time lag

Diffusion analysis outputs:

• Diffusion coefficient (D)
• Diffusion mode (normal, sub-, or super-diffusion)
• Track statistics (number of tracks, track lengths)

Use cases:

• Molecular diffusion in membranes
• Protein mobility in cells
• Nanoparticle diffusion
• Drug molecule tracking
• Lipid raft dynamics
• Receptor diffusion studies

Comparing the three approaches

When to use each workflow

Parameter comparison

Workflow complexity

Simplest: Track moving objects - straightforward tracking + measurements

Moderate: Particle motion - adds motion feature calculations and classification

Advanced: Single particle MSD - requires careful acquisition and specialized analysis

Common tracking challenges

Track fragmentation

Problem: Tracks break into multiple segments instead of continuous trajectories.

Causes:

• Objects disappearing temporarily (out of focus, low signal)
• Detection failures in some frames
• maxGap too small

Solutions:

• Increase maxGap parameter (allow longer disappearance)
• Improve segmentation quality
• Optimize acquisition (better focus, higher frame rate)

Incorrect track assignment

Problem: Tracks switch between objects (ID swapping).

Causes:

• Objects too close together
• maxSpeed too large (allows impossible jumps)
• Similar object appearance

Solutions:

• Reduce object density (lower seeding)
• Decrease maxSpeed for more conservative linking
• Use object shape information (Track Objects instead of Track Particles)

Too many short tracks

Problem: Most tracks only last a few frames.

Causes:

• High object turnover (appearing/disappearing)
• Poor segmentation consistency
• Photobleaching

Solutions:

• Increase MinSegmentCount to filter short tracks
• Improve segmentation robustness
• Reduce imaging intensity to minimize photobleaching
• For MSD: reduce required MinSegmentCount if necessary, but interpret results cautiously

Missing tracks at image boundaries

Problem: Tracks end when objects reach image edges.

Causes:

• Objects leaving field of view
• Border effects in tracking

Solutions:

• Use Remove Touching Frame to exclude border objects
• Acquire larger field of view
• Track only central region if migration is non-directional

Best practices

Experimental design

Imaging parameters:

• Frame rate: Fast enough to capture motion (maxSpeed/2 is a guideline)
• Total duration: Long enough for meaningful statistics (10-20x typical track length)
• Field of view: Large enough to capture full trajectories
• Z-position: Maintain focus (use autofocus for long acquisitions)

Sample preparation:

• Density: Moderate density (avoid overcrowding)
• Labeling: Bright, stable fluorophores
• Media: Minimize drift and evaporation

Tracking parameter optimization

Iterative approach:

1. Start with default parameters
2. Visualize tracks overlaid on images
3. Identify problems (fragmentation, swapping, missing tracks)
4. Adjust parameters accordingly
5. Re-run and verify improvement

Key parameters to tune:

• maxSpeed: Set to 1.5-2× maximum expected speed
• maxGap: Start with 2, increase if detection is inconsistent
• MinSegmentCount: Balance between statistics and data retention

Quality control

Verify tracking quality:

• Visual inspection of trajectories on images
• Check track length distribution (should have a clear peak)
• Monitor track start/end events (should not all be at boundaries)
• Compare forward and backward tracking (should give similar results)

For MSD analysis:

• Examine individual track MSDs before averaging
• Verify linearity in the initial region
• Check that enough tracks contribute to each time lag
• Compare D values across replicates for consistency

Advanced topics

3D tracking

All tracking methods extend to 3D with appropriate inputs:

• Use 3D segmentation (Bright Spots 3D, Threshold 3D)
• Track Particles works with X, Y, Z coordinates
• MSD formula becomes: MSD(Δt) = 6D·Δt (3D motion)

Multi-color tracking

Track particles in multiple channels simultaneously:

• Segment each channel separately
• Track each population independently
• Analyze interactions between populations

Tracking with division

For cells that divide:

• Use specialized cell tracking nodes
• Track lineage relationships (mother-daughter)
• Analyze division timing and frequency

Confined diffusion analysis

For MSD curves showing plateaus:

• Fit confined diffusion models
• Extract confinement radius
• Determine confinement strength

Exporting and further analysis

Results can be exported for external analysis:

Track data:

• Full trajectory tables (X, Y, time, track ID)
• Track summary statistics
• Motion feature tables

For publications:

• High-resolution trajectory images
• MSD curves with error bars
• Statistical comparisons between conditions

Troubleshooting summary

Related workflows

• Object counting - Basic object detection
• Time measurement - Temporal measurements without tracking
• Cell measurement - Cell morphology analysis
• 3D counting and tracking - Extension into 3D