Frangi Vesselness Filter
Detects tubular structures using the Hessian matrix eigenvalues at multiple scales (sigmas). Outputs a probability map where bright = vessel-like. Only sigmas above the selected threshold are used, so higher sigma = only thick vessels detected.
✏ Draw ROI (Region of Interest)
Click to place polygon vertices on the vesselness map. Double-click or snap to the first point to close. Only pixels inside the ROI contribute to the vessel area calculation. Use Copy ROI / Paste ROI to apply the same region across multiple images.
🖌 Paint Tool
Manually mark vessel gaps that the filter missed. Strokes are stored in image coordinates and added to the vessel binary mask when ▶ Run is clicked. Five brush sizes available. Use 🧹 Erase to remove individual strokes.
🎬 Sequence Analysis
For cardiac cycle image stacks. Select frames → compute a projection (Mean/Median/Min/Max Vesselness/Std Dev) → Auto-Optimize min segment size on the projection → Apply Projection Template to all frames → Analyze all frames in batch.
Auto-Optimize
Tests all min segment sizes on the projection image and finds the "elbow" — the point where increasing min size stops removing noise and starts removing real vessels. The bar chart visualizes this transition.
Projection-Guided Fill
For cardiac sequences: where the max vesselness projection detects a vessel (across all frames) AND the current frame has any vesselness signal there (even weak), the threshold is relaxed. Controlled by the Guide Strength slider (0–100%).
📐 Projection Template
Harder version of guided fill. Builds a binary vessel mask from the projection and overlays it as a cyan region on every frame. When ▶ Run is pressed, the template forces vessel detection at 25% of the normal threshold inside template regions.