Count cells or nuclei categorized by a shape parameter in tiled images

Parameters:

Name	#	Type	Description
Selected objects have ... the classifiers below	1	string	Sets up criteria for minimal or maximal values
Classifier: Volume (pixels, 0 for not checking)	2	integer	Volume in 3D, area in 2D. 0 for not checking
Classifier: Perimeter (pixels, 0 for not checking)	3	integer	Perimeter of the maximal cross section in 3D, perimeter in 2D 0 for not checking
Classifier: Diameter (pixels, 0 for not checking)	4	real	Farthest points of the maximal cross section. 0 for not checking
Classifier: Shape factor (0-1, 0 for not checking)	5	real	1 for disc, smaller for irregular shapes. 0 for not checking
Classifier: Fiber Length (pixels, 0 for not checking)	6	real	Provides estimate of length of filamentous objects. 0 for not checking
Classifier: Fiber Breadth (pixels, 0 for not checking)	7	real	Provides estimate of widths of filamentous objects. 0 for not checking
Classifier: Distance to nearest neighbor (pixels, 0 for not checking)	8	real	Calculates distances of centers of gravities of objects. 0 for not checking
Classifier: Number of neighbors (0 for not checking)	9	integer	Neighbors are defined as objects with a distance of center of gravities less than the double of the sum of their diameters. Enter the desired min or max number of neighbors plus one or 0 for not checking
Constrain to ROIs	10	boolean	Calculates only segments overlapping (any small intersection) with the active ROI in the segmented image.
Number of tiles in x	11	integer	The image consists of this number of equal sized tiles in x dimension.
Number of tiles in y	12	integer	The image consists of this number of equal sized tiles in y dimension.
Minimum nucleus diameter	13	integer	Explicit minimum diameter of objects in pixels.
Approximate nucleus diameter	14	integer	Diameter of the nucleus in pixels. Objects varying around this size will be detected.
Background cutoff (%)	15	real	Details dimmer than this in filtered rescaled images will be rejected. Increase this value if debris dimmer than the cells is detected.
Debris cutoff (percentile)	16	real	Brighter details of the image than this will be ignored. Decrease this value if bright debris results wrong scaling of nuclei.
Minimum nucleus fluorescence (%)	17	real	Cells dimmer than this in filtered rescaled images will be rejected. Increase this value if debris dimmer than the cells is detected.
Nucleus boundaries (% of peak fluorescence)	18	real	For "Bound locally", the boundary of each object is determined at this % of the maximal intensity of the object relative to its neighborhood. For "Bound uniformly" this is a pixel intensity value.
Weld segments into round objects	19	boolean	Weld touching segments if they form a rounder object together. Use this to avoid objects fragmenting into multiple segments.

Description:

Count total and a subset of cells or nuclei, categorized by a shape parameter. Set Min or Max in “Selected objects have ... the classifiers below” to specify whether the subset of objects have larger (or equal) or smaller (or equal) values than the classifier set below, respectively. Classifiers with zero value are not checked.
Output: data column Ch1 is the total count, Ch2 is the subset count.

The pipeline removes uneven background and tiling pattern. Then cells or nuclei are identified using watershed segmentation. Finally, using a copy of the segmented image object rejected by the selection classifier are removed. The numbers of objects are counted in each image, either in whole image or in ROIs if present.

Paraformaldehyde fix and stain samples with Hoechst 33342 or DAPI. Tiled image the bottom of the well at low magnification, e.g. 10x (~1.6um/pixel resolution).
Set approximate cell diameter. To this end load an image, zoom in using the magnifier glass Main Toolbar button, and then using the linear ROI button draw a line across a nucleus. Double-click the status bar of the Image Window to see the length of the ROI (Size of active ROI).
Set minimum size threshold.
No other shape classifiers are applied here, e.g. no minimum shape factor in contrast to other cell counting pipelines.
Run the pipeline on a single well and observe the results. In the overlay Image Window the gray fluorescence image should be well matched by the colored segments:
*If single nuclei are detected as multiple segments, increase the approximate cell diameter. In addition, you may turn segment welding on.
*If multiple nuclei are detected as single segments, decrease the approximate cell diameter. In addition, you may turn segment welding off.
*If debris is detected a nuclei, increase background cutoff or minimum cell fluorescence.
*If dimmer cells are missed, decrease background cutoff or minimum cell fluorescence.
*If bright debris outshines cells, decrease debris cutoff percentile.
If segmentation goes as it is desired, process the whole microplate using the double blue arrowhead button and ‘Run Pipeline … on All Stage Positions’.
Tiled images: the background tiling pattern is efficiently removed by spatial filtering if the recording was performed without overlap and image registration. Provide the number of tiles in x and y direction.

Based on "Seahorse well cell count with nuclear stain" V2

Version history:
V2:
Changed shape factor to follow changes in core calculations
Added option to discard segments touching the edges of the image