Measurements Reference

This page provides a comprehensive reference for all single-cell measurements available in Celldetective. The output table (e.g., trajectories.csv or tracked_data.csv) contains one row per cell per time point, with columns corresponding to these measurements.

Note

Channel Naming: For multi-channel images, intensity-based measurements are prefixed with the channel name (e.g., GFP_mean, RFP_max).

Morphological Measurements

Standard shape descriptors computed from the cell’s segmentation mask using scikit-image.regionprops.

Measurement Name	Description
`area`	Number of pixels in the region.
`perimeter`	Perimeter of object which approximates the contour as a line through the centers of border pixels using a 4-connectivity.
`eccentricity`	Eccentricity of the ellipse that has the same second-moments as the region. The eccentricity is the ratio of the focal distance (distance between focal points) over the major axis length. The value is between 0 and 1.
`solidity`	Ratio of pixels in the region to pixels of the convex hull image.
`extent`	Ratio of pixels in the region to pixels in the total bounding box.
`orientation`	Angle between the 0th axis (rows) and the major axis of the ellipse that has the same second moments as the region (in radians).
`axis_major_length`	The length of the major axis of the ellipse that has the same normalized second central moments as the region.
`axis_minor_length`	The length of the minor axis of the ellipse that has the same normalized second central moments as the region.
`equivalent_diameter_area`	The diameter of a circle with the same area as the region.
`feret_diameter_max`	Maximum Feret diameter (longest distance between any two points on the region contour).
`euler_number`	Euler characteristic of the set of non-zero pixels. Computed as number of objects minus number of holes (using 8-connectivity).
`convex_area`	Number of pixels of convex hull image, which is the smallest convex polygon that encloses the region.
`centroid-0`, `centroid-1`	Coordinate of the centroid (row, col). Note: In tracked data, these are replaced by `POSITION_Y` and `POSITION_X`.

Intensity Measurements

Statistics of pixel intensities within the cell mask.

Measurement Name	Description
`{channel}_mean`	Mean intensity. NaN values are ignored.
`{channel}_max`	Maximum intensity.
`{channel}_min`	Minimum intensity.
`{channel}_std`	Standard deviation of intensity.

Center of Mass Measurements

Quantifies the spatial distribution of intensity relative to the geometric center.

Measurement Name	Description
`{channel}_center_of_mass_distance`	Euclidean shift between geometric centroid and intensity center of mass.
`{channel}_center_of_mass_angle`	Angle of the shift vector (radians).
`{channel}_center_of_mass_dx`	Shift along X-axis (pixels).
`{channel}_center_of_mass_dy`	Shift along Y-axis (pixels).

Radial Intensity Measurements

Analysis of intensity distribution from center to periphery.

Measurement Name	Description
`{channel}_radial_intensity_slope`	Slope of linear regression (intensity vs radius).
`{channel}_radial_intensity_intercept`	Intercept at center.
`{channel}_radial_intensity_r2`	$R^2$ goodness-of-fit to linear gradient model.

Texture Measurements (Haralick)

Note

Texture features are computed using the Haralick method (via the mahotas library). These features are calculated on a per-channel basis.

Naming Convention:

haralick_{feature}_{channel}

where {feature} is one of the following 13 Haralick features:

angular_second_moment
contrast
correlation
sum_of_square_variance
inverse_difference_moment
sum_average
sum_variance
sum_entropy
entropy
difference_variance
difference_entropy
information_measure_of_correlation_1
information_measure_of_correlation_2
maximal_correlation_coefficient

Example: For a channel named EGFP, the contrast would be: haralick_contrast_EGFP.

Tip

Haralick features are computed using a specific distance (offset) defined in the settings. However, the distance is not included in the feature name. If you process multiple distances, keep in mind they might overwrite each other if output to the same table.

Spot Detection

Note

Spot detection counts “blobs” or spots within the cell area (e.g., vesicles, granules) using LoG (Laplacian of Gaussian) or DoG (Difference of Gaussian).

Naming Convention:

{channel}_spot_count: Total number of spots detected in the cell for the given channel.
{channel}_mean_spot_intensity: Mean intensity of the detected spots (pixels belonging to spots) for the given channel.

Example: For a channel named Lysosomes, you will get:

Lysosomes_spot_count
Lysosomes_mean_spot_intensity

Extra Properties (Custom)

Celldetective includes several built-in “extra properties” that provide more advanced morphological or intensity descriptors.

Area/Fraction in Intensity Threshold:

area_detected_in_{channel}: Area (in pixels) of the cell that is above a certain intensity threshold in {channel}.
fraction_of_area_detected_in_{channel}: Fraction of the cell’s total area that is above the threshold (0.0 to 1.0).
area_dark_in_{channel}: Area (in pixels) of the cell that is below a certain intensity threshold.
fraction_of_area_dark_in_{channel}: Fraction of the cell’s total area that is below the threshold.

Other Custom Properties:

intensity_nanmean: Mean intensity ignoring NaNs (useful for masked arrays).
feret_diameter_max: Maximum Feret diameter.

Neighborhood Measurements

Neighborhood measurements describe the local environment of a cell, quantifying its neighbors. These are often computed between two populations (e.g., Targets vs Effectors) or within a single population.

Naming Convention:

{metric}_count_neighborhood_({pop1}-{pop2})_{method}_{radius}_px

Where:

{metric} is one of:
- inclusive: Total number of neighbors.
- exclusive: Number of neighbors for which the current reference cell is the absolute closest.
- intermediate: A weighted count sum of weights, where $w = 1/N_{neighborhoods}$. A neighbor belonging to 2 references contributes 0.5 to each.
{pop1}-{pop2}: The pair of populations interacting (e.g., targets-effectors).
{method}: The neighborhood definition method:
- circle: Neighbors within a fixed Euclidean distance.
- contact: Neighbors touching or within a small dilation distance (mask-based).
{radius}: The search radius or duality distance in pixels.

Example: inclusive_count_neighborhood_(targets-effectors)_circle_30_px

Status Decomposition:

If the analysis distinguishes between neighbor states (e.g., “live” vs “dead”):

_s1_: Positive status (e.g., live neighbors).
- Example: inclusive_count_s1_neighborhood_(targets-effectors)_circle_30_px
_s0_: Negative status (e.g., dead/no-event neighbors).
- Example: inclusive_count_s0_neighborhood_(targets-effectors)_circle_30_px

Mean Neighborhood Relative to Events:

When analyzing tracks with detected events (e.g., cell division, death, or a specific signal change), neighborhood metrics can be aggregated relative to the event time ($t_{event}$) of the reference cell.

mean_count_{metric}_{neigh_col}_before_event
- Description: The average of the {metric} (e.g., inclusive, intermediate) calculated over all frames up to and including the event frame of the reference cell ($t le t_{event}$).
- Note: If no event is detected for the reference cell (or no event column is specified), this metric is calculated over the entire duration of the track ($t le t_{max}$).
- Implicit Status: This metric specifically averages the counts of neighbors with positive status (_s1_), e.g., live cells.
mean_count_{metric}_{neigh_col}_after_event
- Description: The average of the {metric} calculated over all frames strictly after the event frame of the reference cell ($t > t_{event}$).
- Note: If no event is detected for the reference cell, or if the event occurs at the last frame, this value will not be computed (NaN).
- Implicit Status: This metric specifically averages the counts of neighbors with positive status (_s1_).

For details on neighborhood analysis, see How to measure cell-cell interactions.