- Acquire high-resolution time-lapse images of the hindgut.
- Crop raw time-lapses to zoom in on the hindgut region (using ImageJ).
- Remove transition frames (e.g., due to denervation or drug application).
- Downsample from 30 to 3 fps to match the long timescales of gut movement.
- Temporally smooth movies using Simoncelli filter (5-point weights).
- Avoid spatial smoothing to retain features critical for flow analysis.
- Hindgut Segmentation: Use MATLAB's
strelandimerodeto segment gut outlines frame-by-frame. - Optical Flow Analysis: Apply the Lucas-Kanade method using a Gaussian-weighted neighborhood (σ = 2 px).
- Impose a reliability threshold (0.01) to remove unreliable vectors.
- Set flow vectors below the threshold to zero.
- Dorsal-Ventral Motility Map: Average lateral flow components across the dorsal-ventral axis to generate motility maps.
- Flow Smoothing for Visualization: Apply Gaussian smoothing to visualize motion direction and reduce localized noise.
- Apply FFT along the time axis to compute 2D power spectral density (PSD).
- Extract dominant motion frequencies and harmonics.
- Nyquist frequency is 1.5 Hz (sampling at 3 Hz).
- Calculate power as the sum of squared FFT amplitudes normalized by signal length.
- Also compute instantaneous power using FFT along the spatial axis.
- Define Relative Rhythmic Power as the power in the peak frequency band (±0.04 Hz) divided by total power.
- Used to quantify signal coordination:
- White noise → flat PSD → low rhythmic power.
- Single-frequency signal → narrow PSD peak → high rhythmic power (~1).
- Metric is robust to small changes in frequency window width.
- Apply speed threshold (top 30% globally) to highlight wave regions.
- Use MATLAB's
bwareaopenandbwlabelnto find connected wave regions (area > 100 px). - Extract waves that span >1/5 of gut length.
- Use
bwskelto trace wave skeletons and compute speed via slope. - Label wave direction:
- AP (anterior → posterior): positive speed
- PA (posterior → anterior): negative speed
- Mixed: zero speed
- For each phase:
- Use last 15 minutes of Phases 1 & 2
- Use first 15 minutes of Phase 3
- Rationale:
- Capture short-term changes following 5-HT application.
- Supported by statistical testing between 15-min windows.
- Statistical Analysis:
- Compare motility features across conditions (e.g., saline vs. N7 cut vs. serotonin).
- Use non-parametric tests (e.g., Wilcoxon signed-rank).
- Visualization:
- Motility kymographs, PSD plots, and boxplots.
- Color-coded vector fields for flow direction and magnitude.
- Image Processing: Cropping, thresholding, morphological operations.
- Flow Estimation: Lucas-Kanade method with Gaussian weighting.
- Signal Processing: FFT-based PSD, rhythmic power analysis.
- Wave Extraction: Connected component analysis and wave skeleton tracing.
- Statistical Analysis: MATLAB functions (
signrank,boxplot) for paired comparison. - Software: MATLAB, ImageJ.