I have more than 10k images (Phase contrast, 10X) of hMSCs and I need to count the cells. Right now, I am using ImagaeJ jo manually count them in a 'tick-tick' fashion which is very laborious and time-consuming. As there is not a very sharp contrast between the cells and the background, can a MATLAb code be written to ease my work?
Hello Moin,
I remember there was a method in Image J to convert the images to binary image and using the threshold feature distinguish cells from background in an automated fashion. Please have look to his links which might be helpful
https://www.unige.ch/medecine/bioimaging/files/3714/1208/5964/CellCounting.pdf
https://bitesizebio.com/30184/quick-easy-automatic-cell-counting/
In Matlab you could use a usual threshold such as Otsu, then use bwlabel to get different regions contaning a cell. problem would be characterising what is and what is not a cell because your illumination is not quite homogeneous. But these kind of countign algorithms are numerous.
Also if you don't have a good contrast you may consider using gradient to simulate contrast on your image.
There is this tutorial in Python, which can easily be exported to matlab, for segmentation and counting :
http://scikit-image.org/docs/dev/user_guide/tutorial_segmentation.html
To count easilly cells, I usually stain my cells with dapi at 0.05µg/ml and count using Image J the stained nuclei. Fluorescent images were transformed in binary using the tab Process/Binary/Make binary. Segmentation will be done using the watershed option (under the tab Process/Binary/Watershed).The nuclei were counted under the tab Analyze/Analyse particles. You need to modify the particle size to avoid counting small particles that were not nuclei, usually I use the particle size at 40-10000 pix². I hope this will help you.
Yes, you may watch https://www.youtube.com/watch?v=D1qBaFwuF4E
https://www.unige.ch/medecine/bioimaging/files/3714/1208/5964/CellCounting.pdf
- Badges
- Science method