Hello all, I am a beginner trying to use Ilastik for one of our research projects. I am running into an issue where the output images from Ilastik are showing all predicted objects, even those with a low probability of being a cell. This is an issue because we are trying to perform extra analysis on these outputted images in ImageJ, and we only want to analyze the objects with a high probability of being a cell. I have tried formatting the output image to output object probabilities, which gives me a color output with the color corresponding to the probability of the object being a cell (red means high probability, green means low probability). From this, I am splitting the color channels in ImageJ and analyzing the red color channel. However, on the cells that have an “in-between” probability, still appear in the red color channel. This would be ok if I knew what the actual probability was for that specific object. Is there a way to either have Ilastik only output the objects considered cells, or a better way to handle this is ImageJ?
Hi, in Zen Blue Intellesis there exists the posibility to apply an propability threshed to the segmented objects in order to only perform the subsequent measurements using pixels above a certain probability threshold.
Only the objects the “survive” will be used for measurements.
It can be used on any image type.
Remark: This a commercial software with a free trial tocheck it out.
Thank you for the suggestion, I will give it a try.
which ilastik workflow are you using? Pixel Classification or Object Classification?