Computation of cell morphological features

[PierreBSC]

Hi,

I am currently using the ImcSegmentationPipeline on the test data provided and I just realised that the last step of the pipeline does not export/produce by default classical cell morphological features such as the size of the cell, the eccentricity, granularity etc.. Is there currently anyway to easily get such features with the pipeline ?

Thank you in advance for your help :)

Best

Pierre

[votti]

Hi Pierre, This is indeed not currently included: In IMC the resolution is rather low (usually cuts through cells of >4um thickness acquired at 1x1um lateral resolution), thus I have rarely seen any morphological feature besides Area been productively used.

If you think they are useful for your project, you can add the 'measureobjectsizeshape' module (1) to the '3_measure_mask_basic' pipeline (2) to get these measurements. It should be straight forward to do this by opening the pipeline in Cellprofiler.

Best, Vito

(1) http://cellprofiler-manual.s3.amazonaws.com/CellProfiler-3.0.0/modules/measurement.html#measureobjectsizeshape

(2) https://github.com/BodenmillerGroup/ImcSegmentationPipeline/blob/main/cp4_pipelines/3_measure_mask_basic.cppipe

[nilseling]

Hey @votti, wouldn't it be good to add the Area measurement in the default pipeline? For the downstream workflow I tend to visualize this for QC purposes. Cheers, Nils

[votti]

I am quite impartial about this. The consensus so far was to only measure by default what is effectively used by the usual downstream processing.

Thus morphological, subcellular compartments and texture features were so far not included in the basic pipeline.

If you find value in it these measures they can easily be added. Please open a pull request.

A side note: A detail about these features is that most of them need different downstream processing (eg transformations) than the intensity features, if used for analysis.

Cheers!

On Thu, Jan 28, 2021, 20:53 Nils Eling notifications@github.com wrote:

Hey @votti https://github.com/votti, wouldn't it be good to add the Area measurement in the default pipeline? For the downstream workflow I tend to visualize this for QC purposes. Cheers, Nils

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/BodenmillerGroup/ImcSegmentationPipeline/issues/56#issuecomment-769337058, or unsubscribe https://github.com/notifications/unsubscribe-auth/AA4R2KF3SGUEJL3CRIVG6CTS4G6EBANCNFSM4WXG5MIA .

[plankter]

As a workaround, in Python one can use regionprops or regionprops_table methods of scikit-image library to measure these properties passing a mask array: https://scikit-image.org/docs/dev/api/skimage.measure.html#skimage.measure.regionprops

Best regards, Anton

[PierreBSC]

Hi Vitto, Anton and Nils,

Thanks for the very detailed response. I will try to add the module you suggested and digg into the literature to see if some basic morphological features could not be useful for QC and basic cell type classification.

Best

Pierre

[PierreBSC]

Ok so after few technicals issues I modified the 3_measure_mask_basic so that it exports the Area features of the cells. In addition I have also removed some of the Image intensity features to make the Cell file lighter. Will probably make a few more modification before adding it to the GitHub next week !

Best

Pierre

[nilseling]

Morphological features are now being exported by default via #87