Identification of cells in muscle biopsy image


Our group is trying to use the software to automate the process of quantifying the intensity of individual cells in a muscle biopsy which have undergone staining. However, we are having real problems tweaking the module ‘Identify Primary Objects’ to select the cells. We’ve tried many different combinations of settings and the best we can do is not very good (see attached images).

The best one (attached) used the following settings: ImageMath to invert. IdentifyPrimaryObjects with: 15-60 diameter, Otsu Adaptive, Three Class Thresholding (middle = foreground), Boundaries on Thresholding = 0.0 - 0.9, Clumping by Shape and Lines by Shape.

As we are new to the software, I wondered if there is something obvious that we are missing? Which setting is the most likely culprit?


University of Newcastle, UK

Hi reh1,

I’m attaching a pipeline which attempts to perform this task, although the contrast is extremely low (as you note), so additional adjustment of the settings in IdentifyPrimaryObjects is needed. Essentially, the pipeline consists of attempting to identify the tissue as a whole (straightforward) and then the individual cells (not as straightforward).

2011_10_06.cp (6.17 KB)