Segmentation

The Segmentation tab provides tools for refinement, clustering and analysis.

Merge

#merge

Remove

#remove

Select

#select-by-size

Transform

#transform

Crop

#crop

Cluster

#cluster

Outlier

#outlier_removal

Normals

#normals

Trim

#trim

Skeletonize

#skeletonize

Downsample

#downsample

Properties

#properties

Merge

Combines multiple clusters or creates new clusters from point selections:

For complete clusters:

1. Select multiple clusters in the Object Browser

2. Click Merge in the ribbon or press m after clicking the viewport.

3. Selected clusters are combined into a single new cluster

For point selections:

1. Use area selection (R key) to select points from one or more clusters

2. Click Merge or press m after clicking the viewport.

3. A new cluster is created containing only the selected points

4. Original clusters remain but without the selected points

Remove

Deletes selected clusters or removes points from clusters:

For complete clusters:

1. Select one or more clusters in the Object Browser

2. Click Remove or press Delete after clicking the viewport.

3. Selected clusters are completely deleted

For point selections:

1. Use area selection (R key) to select points within clusters

2. Click Remove or press Delete after clicking the viewport.

3. Only the selected points are removed from their parent clusters

4. Empty clusters are automatically deleted

Select by Size

Filters clusters by point count:

1. Click Select in the ribbon

2. Adjust the slider to set a minimum size threshold

3. Clusters below the threshold are automatically selected

4. Use in combination with Remove to clean up small clusters

Transform

Applies rotation and translation to clusters:

1. Select a cluster in the Object Browser

2. Click Transform

3. A 3D transformation widget appears around the cluster

4. Use the transformation widget to move or rotate the cluster

5. Press Transform again to exit transformation mode

Crop

Trims points based on distance to other structures:

1. Click Crop

2. Select source structures to crop

3. Select target structures to measure distance from

4. Set the distance threshold

5. Choose to keep points within or beyond the threshold

Cluster

Groups points into separate clusters:

1. Select a cluster with multiple distinct structures

2. Click Cluster

3. Choose clustering method:

   • Connected Components: Groups connected components (default). Particularly useful for postprocessing volume segmentations

   • Envelope: Retrieve boundaries of dense membrane segmentation

   • Leiden: Partition connected segmentations into distinct objects

   • DBSCAN: Density-based clustering with distance and minimum points parameters

   • K-Means: Divides into a specified number of clusters

   • Birch: Hierarchical clustering

4. Configure method-specific parameters:

   Leiden:

   • Resolution: Clustering resolution. Lower values yield larger cluster.

   DBSCAN:

   • Distance: Maximum distance between points in the same cluster

   • Min Points: Minimum points required to form a cluster

   K-Means:

   • K: Number of target clusters

   Birch:

   • Clusters: Number of target clusters

   • Threshold: Radius threshold for merging subclusters (lower values create more clusters)

   • Branching Factor: Maximum subclusters per node (affects memory usage and clustering speed)

5. Click OK to apply clustering

Outlier Removal

Removes noise points using statistical methods:

1. Select a cluster to clean

2. Click Outlier

3. Choose removal method:

   • Statistical: Removes points based on distance to neighbors

   • Eigenvalue: Removes edge points using covariance analysis

4. Configure method-specific parameters:

   • Neighbors: Number of neighbors to consider for statistics

   • Threshold: Sensitivity of outlier detection (lower = more aggressive)

Normals

Modulate normals of a point cloud object

1. Select a cluster

2. Click Normals

3. Choose method:

   • Compute: Recompute normals by orienting point cloud normal vector field.

   • Flip: Flip normals

4. Configure method-specific parameters:

   Compute:

   • Neighbors: Number of neighboring points to consider

Trim

Select points outside specified axis-aligned boundaries:

1. Select a cluster

2. Click Trim

3. Two cutting planes appear in the 3D viewer

4. Position the planes by dragging or use keyboard shortcuts:

   • X: Align planes to X-axis

   • C: Align planes to Y-axis

   • Z: Align planes to Z-axis

5. Points between the planes are preserved

6. Press Trim again to exit trim mode

Skeletonize

Skeletonize point cloud:

1. Select a cluster

2. Click Skeletonize

3. Choose method:

   • Core: Classical internal skeleton.

   • Boundary: Exo-like skeleton of boundaries.

   • Outer: Outer exo-like skeleton.

   • Outer_Hull: Legacy method to compute the outer hull of a point cloud. Used to be listed unter Thin pre v1.0.16.

4. Click OK to apply thinning

Downsample

Reduces the number of points while maintaining overall structure:

1. Select a cluster

2. Click Downsample

3. Choose downsampling method:

   • Radius: Remove points within a specified distance of each other

   • Number: Randomly subsample to a target number of points

4. Configure parameters:

   • Radius: Minimum distance between retained points

   • Size: Target number of points for random subsampling

5. Click OK to apply downsampling

Properties

Advanced analysis and visualization dialog with three modes: Visualize, Distribution, and Statistics.

1. Select objects in the Object Browser

2. Click Properties in the ribbon

3. Use Compute to calculate properties, then switch between tabs

Property Categories:

• Distance: To camera, clusters, or models

• Surface: Curvature, edge length, surface area, volume

• Geometric: Dimensions, point counts, identity

• Projection: Projected curvature, geodesic distance

Visualization Options:

• Color Maps: Common colormaps (viridis, plasma, etc.)

• Normalization: Per-object or global scaling

• Quantiles: Statistical binning for outlier handling

• Interactive: Real-time color mapping in 3D viewport

Visualize Tab: Compute geometric properties and display as interactive color maps in the 3D viewport.

Distribution Tab: Generate interactive export-ready histograms, density plots, and line charts with customizable styling.

Statistics Tab: View numerical summaries (min, max, mean, std dev) and export data as CSV/TSV files.

Tip

All data can be exported using the Export Data button.