URGENT Question regarding Percent of Area%.
I need to measure the amount of positive (DAB positive) in IHC pictures. After I assign class positive and class negative (the nuclei), I need to select background in order to count the amount of negative and positive pixels. How is the percent of area calculated? what it is Area (pixel^2)? I need to understand how the background area is influencing the percent of positive and how should I take into account the negative area for the analysis. thanks please let me know as soon as possible
0
Best Answer
-
Hi Valentina,
Ok, if you just want to measure total Area of your staining, then it's reported by the Sum statistical parameter of Area measurement. The Area is measured in calibrated units, if you image has calibration, otherwise in pixels. You can use grouping by Class and activate Sum as Totals to see Sum values for both classes.
To check relative area of staining on the image, you can use Percent Area(%) (100% is whole image), it can also be reported as Sum per class.
Yuri0
Answers
1. Percent area (%) and Percent area ratio - use total ROI area.
2. Percent area parent - uses the area of parent object.
Parents can be created using Class 1 and then you can count Nuclei using Class 2 or higher. Use "Preserve existing data" option in the Count options. "Parent Name" will report the name of the parent object and "Percent area parent" will report the ratio of the object area to parent area.
Check some other links related to Parent/Child measurements:
http://forums.mediacy.com/discussion/comment/964/#Comment_964
http://www.mediacy.com/index.aspx?page=ViewSolution&solid=1732
http://www.mediacy.com/index.aspx?page=IP_Premier_Colocalization
Regards,
Yuri
this is not my question I think... I am not using parent/child measurements. I have assigned class negative and class positive and background. Then I press count and I get the data table with different parameters (percent of area for both classes for example).
Is that calculated by %positive area=positive pixels area/(background pixels+positive pixels area+negative pixels area) ?
What do you actually want to measure?
Yuri
In order to be precise and better distinguish brown from blue (which is nuclear staining) as my protein can be both non nuclear and nuclear, I was suggested to assign 2 classes plus the background (I used the SMART tool). The assignments of the 2 classes are pretty accurate actually, but Now I don't know what parameter to use from the data table.
valentina
Or?
In general, with Threshold segmentation, classes can overlap, also you could have a separate segmentation class "Background" which is not the whole image, so Class1 + Class2 +Background may not give the total image area.
i was following your advice:
But i do not see my object area ratio to parent area (class1)
the ratio is reported to whole immage.
As I can see from the screenshot that you don't have Class 1 parents.
The Count must be done in 2 steps:
1. Select Class 1 (or reset Classes), switch off "Preserve existing data". Segment parent, click Count. - you will get your parents.
2. Activate "Preserve existing data", "Use as outlines", segment children, click Count - you will get correct parent/child associations and measurements.
Regards,
Yuri
Yuri
Yuri
i noticed later that it is buggy. if i add 2 child classes it counts everything well.
if i add 3 child classes it does not count..:((
If you still can't find the problem, send me the test image and segmentation ranges (saved from the Threshold tool) and I will check them.
Yuri
After restartin softweare began to work.
So if this simple solution works i do not see any benefits of using Child/Parent aplication from now.
thanks
The app provides a workflow to segment parent and multiple child classes using different segmentation options for each parent and child class, so it can handle more complex cases.
Yuri