How to measure single-cell texture
This guide shows you how to measure Haralick Texture Features on a per-cell basis.
Reference keys: texture features, single-cell measurement
Prerequisite: You must have segmented the cells. Tracking is recommended but not required.
Enable Haralick Texture Features
In the block of your population of interest, click the button of the MEASURE row to open the measurement settings.
In the MASK-BASED MEASUREMENTS section, tick Measure Haralick texture features.
Configure the parameters
Texture options. The Haralick texture options of the measurement settings.
Once the option is ticked (1), the two buttons on its right (2) help choose the parameters on the current position: plots the intensity histogram of the target channel, to check which values the normalization clips, and shows the image digitized to the chosen number of gray levels.
Target channel: select the channel to analyze (3), e.g. a DNA channel for chromatin texture.
Distance, # gray levels and Scale (4):
Distance: the pixel distance for the gray-level co-occurrence matrix computation (default:
1). Larger values capture coarser texture patterns.# gray levels: the number of quantized gray-level bins (default:
256). Lowering this value (e.g.,64) significantly speeds up computation at the cost of intensity resolution.Scale: a downscaling factor between
0and1to reduce the image size before the computation. Useful for large cells.
Normalization (5): intensities are clipped before quantization, between a Min percentile and a Max percentile (e.g., 0.01% – 99.9%). The button (6) switches to absolute mode, where you set a Min value and a Max value in intensity units.
Run the measurements
Scroll down and click Save to save the configuration.
In the control panel, check the MEASURE box and click Submit.
The following Haralick Texture Features will be appended to your measurement table: haralick_contrast, haralick_dissimilarity, haralick_homogeneity, haralick_energy, haralick_correlation, haralick_ASM.
Note
Haralick Texture Features are computationally expensive. Consider lowering the gray levels or using a scale factor < 1 for large datasets.