Morphology¶

Remove Small Obj¶

Removes small connected components from a binary mask.

Details

Objects with area at or below the threshold are discarded. Useful for cleaning noise after thresholding.

max_size — maximum object area in pixels to remove (default: 500).

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Properties: Maximum Size (px²)

Remove Small Holes¶

Fills small enclosed holes in a binary mask up to a given area threshold.

Details

Only holes (background regions fully enclosed by foreground) smaller than the threshold are filled; larger holes remain. Contrast with FillHolesNode which fills all holes regardless of size.

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Properties: Max Hole Size (px²)

Keep Max Intensity¶

Keeps the top N connected components ranked by total intensity in the reference image.

Details

Finds all connected components in the mask, sums each region's pixel intensities from the intensity image input, and retains only the brightest regions as a binary mask.

top_n — number of highest-intensity regions to keep (default: 5).

Direction	Port	Type
Input	`intensity_image`	image
Output	`mask`	mask

Properties: Top N Regions (by area)

Distance Ring Mask¶

Creates an annular (ring) mask by expanding foreground objects outward by a specified distance.

Details

Computes the Euclidean distance transform of the background and selects pixels within the given range, producing a ring-shaped region around the original mask.

local_distance — maximum expansion distance in pixels (default: 200).

Direction	Port	Type
Input	`mask`	mask
Output	`ring_mask`	mask

Properties: Local Distance (px), ``

Erode Mask¶

Applies binary erosion to shrink foreground regions of a mask.

Details

Each iteration removes one layer of boundary pixels. Useful for separating touching objects or removing thin protrusions.

iterations — number of erosion passes (default: 1).

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Properties: Iterations

Dilate Mask¶

Applies binary dilation to expand foreground regions of a mask.

Details

Each iteration adds one layer of boundary pixels. Useful for bridging small gaps or growing regions uniformly.

iterations — number of dilation passes (default: 1).

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Properties: Iterations

Morph Open¶

Applies binary opening (erosion then dilation) to a mask.

Details

Removes small foreground objects and thin protrusions while preserving the overall shape of larger regions.

iterations — number of opening passes (default: 1).

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Properties: Iterations

Morph Close¶

Applies binary closing (dilation then erosion) to a mask.

Details

Fills small holes and connects nearby fragments without significantly changing the overall region size.

iterations — number of closing passes (default: 1).

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Properties: Iterations

Fill Holes¶

Fills all enclosed background holes in a binary mask regardless of size.

Details

Uses scipy.ndimage.binary_fill_holes. Contrast with RemoveSmallHolesNode, which only fills holes smaller than a user-defined area threshold and leaves larger holes intact.

Direction	Port	Type
Input	`mask`	mask
Output	`mask`	mask

Skeletonize¶

Reduces foreground regions to 1-pixel-wide centrelines (skeleton).

Details

Uses skimage.morphology.skeletonize. Outputs SkeletonData, which can feed SkeletonAnalysisNode or any node that accepts a mask.

Method — zhang (default) or lee. Lee tends to produce cleaner skeletons on thick blob-like shapes.
Prune Spurs — iteratively removes endpoint pixels (dead-end tips). Setting N removes all branches shorter than N pixels (0 = off).

Direction	Port	Type
Input	`mask`	mask
Output	`skeleton`	skeleton

Properties: Method, Prune Spurs (iterations)

Skeleton Analysis¶

Analyses a skeleton produced by SkeletonizeNode.

Details

Outputs:

skeleton — filtered SkeletonData (short isolated segments removed)
stats — total skeleton_length_px and junction_count (after filtering)
junction_image — RGB visualisation: skeleton in grey, junction pixels in red on black
junction_mask — binary MaskData of junction pixels only (pass to DrawShapeNode)
Min Segment Length — remove isolated skeleton segments shorter than this many pixels (0 = keep all).
Junction Radius — dilate each junction point into a filled circle of this radius in pixels for the junction_mask output (default: 4).

Direction	Port	Type
Input	`skeleton`	skeleton
Output	`stats`	table
Output	`junction_image`	image
Output	`junction_mask`	mask

Properties: Min Segment Length (px), Skeleton Width (px), Junction Radius (px)

Medial Axis¶

Computes the medial axis (distance-based skeleton) of a binary mask.

Details

Uses skimage.morphology.medial_axis, which finds the set of pixels equidistant from the nearest background pixel. Unlike skeletonize, it also produces a distance transform that encodes the local radius at each skeleton point.

Outputs:

skeleton — the 1-pixel-wide medial-axis skeleton (SkeletonData)
distance — skeleton coloured by local radius using the plasma colourmap (purple = thin, yellow = thick); background is black

Direction	Port	Type
Input	`mask`	mask
Output	`distance`	image

Particle Props¶

Labels connected components in a mask and measures each region.

Details

Shape columns (always present):

label — integer region ID (matches label_image pixel values)
area — number of pixels in the region
equivalent_diameter — diameter of a circle with the same area
centroid_y / centroid_x — pixel coordinates of the region centre
bbox_top / bbox_left / bbox_bottom / bbox_right — tight bounding box corners
perimeter — outer boundary length in pixels
circularity — 4*pi*area/perimeter^2; 1.0 = perfect circle, lower = more irregular
eccentricity — 0 = circle, 1 = line; measures elongation
orientation — angle of major axis in degrees
major_axis / minor_axis — lengths of the fitted ellipse axes
solidity — area / convex_hull_area; 1 = perfectly convex
extent — area / bounding_box_area; fraction of bbox filled
euler_number — 1 = no holes; decreases by 1 for each enclosed hole

Intensity columns (present only when intensity_image is connected):

mean_intensity — average pixel value inside the region
sum_intensity — total pixel intensity (mean x area)
max_intensity / min_intensity — brightest and darkest pixels
weighted_centroid_y / weighted_centroid_x — intensity-weighted centre

Outputs:

table — TableData with all columns above
label_image — LabelData with integer label array and coloured RGB visualisation

Direction	Port	Type
Input	`intensity_image`	image
Output	`table`	table
Output	`label_image`	label_image

Properties: Min Region Area (px²)

Particle Select¶

Filters particles from a LabelData input using dynamic filter rules.

Details

Add filter rules (e.g. area >= 100, circularity >= 0.5) to keep only particles matching all conditions. Available properties are auto-detected from the upstream data.

Direction	Port	Type
Input	`label_image`	label_image
Input	`image`	image
Output	`mask`	mask
Output	`label_image`	label_image

Visual Particle Select¶

Selects particles visually by clicking them directly in the label image.

Details

Displays the label image with coloured regions. Click a particle to deselect it (dims); click again to re-select. All / None buttons for bulk operations. Min/Max area filters to exclude tiny debris.

Same inputs and outputs as Particle Select but with visual, spatial interaction instead of a checkbox list.

Direction	Port	Type
Input	`label_image`	label_image
Input	`image`	image
Output	`mask`	mask
Output	`label_image`	label_image

Particle Classify¶

Assigns particles to named groups by drawing shapes around them.

Details

Usage¶

Connect a label image (from Particle Props or Watershed).
Click + New Group to create groups (Group 1, Group 2, …).
Select a group, then draw a rect or lasso around particles.
All labels fully enclosed by the shape are assigned to that group.
Unassigned labels remain as "Ungrouped" (label 0 in output).

Outputs¶

label_image — relabeled: Group 1 → 1, Group 2 → 2, ungrouped → 0.
table — original particle table with an added group column.

Direction	Port	Type
Input	`label_image`	label_image
Input	`image`	image
Output	`label_image`	label_image
Output	`table`	table

White Top-Hat¶

Applies a morphological white top-hat filter to extract small bright features.

Details

Subtracts the morphological opening (background estimate) from the original image, leaving only bright structures that fit inside the disk of the given radius. Useful for equalising uneven illumination or removing broad bright background before thresholding. Works on grayscale and each channel of RGB independently.

Disk Radius — radius of the structuring element in pixels.

Direction	Port	Type
Input	`image`	image
Output	`image`	image

Properties: Disk Radius (px)

Black Top-Hat¶

Applies a morphological black top-hat filter to extract small dark features.

Details

Subtracts the original image from its morphological closing, revealing small dark structures (holes, valleys, cracks) on a bright background. The complement of White Top-Hat -- use White Top-Hat for bright features on dark backgrounds. Works on grayscale and each channel of RGB independently.

Disk Radius — radius of the structuring element in pixels.

Direction	Port	Type
Input	`image`	image
Output	`image`	image

Properties: Disk Radius (px)

Watershed¶

Separates touching or overlapping objects using marker-controlled watershed segmentation.

Details

Pipeline:

Compute Euclidean distance transform of the binary mask.
Find local maxima (object centres) in the distance map; Min Object Sep. controls the minimum allowed gap between two peaks.
Run watershed on the inverted distance map to delineate each object.

Table columns:

label — integer region ID (matches label_image pixel values)
area — number of pixels in the region
equivalent_diameter — diameter of a circle with the same area (sqrt(4*area/pi))
centroid_y, centroid_x — pixel coordinates of the region centre
perimeter — outer boundary length in pixels
circularity — 4*pi*area/perimeter^2; 1.0 = perfect circle, lower = more irregular
eccentricity — 0 = circle, 1 = line; measures elongation
orientation — angle of major axis in degrees; 0 = right, +90 = up, -90 = down
major_axis — length of longest axis of the fitted ellipse
minor_axis — length of shortest axis of the fitted ellipse
solidity — area / convex_hull_area; 1 = perfectly convex, <1 = concave
extent — area / bounding_box_area; fraction of bounding box filled
euler_number — 1 = no holes; decreases by 1 for each enclosed hole

Direction	Port	Type
Input	`mask`	mask
Output	`label_image`	label_image
Output	`table`	table

Properties: Min Object Sep. (px)