To find generalised threshold value of an image

I extracted the tumour from the original image by using the following code. I found the threshold value by trial and error method. How to find the generalised threshold value?

I=imread('tum.tif');
a=imadjust(I);
a=imadjust(I,[0.3; 0.7],[0.3; 0.7]);
b=im2bw(a);

[Rik]

What do you mean with generalised threshold? If you want to have a method that works for all tumors, I'll be sending my solution to the Nobel Committee instead of posting it here.

Otherwise, kmeans is a good place to start if there is a marked intensity difference between tumor and non-tumor. There are a lot more possibilities. Maybe a better place to start is googling 'intensity based segmentation techniques'.

[rohith bharadwaj]

So does that mean for every image I will have to change the thresholding value accordingly on the basis of trial and error?

[Rik]

That depends on how general you want your code to work. There might not be answer that works for all tumors, but density based segmentation might work for all your images. I'm sure you'll be able to find some examples if you type that into your favorite internet search engine.

Answer 1

[Image Analyst]

I also don't know what a "generalised" threshold is. I disagree that kmeans is a good method. While it may work good for images where you know for a fact that there is definitely a good sized tumor there, it will not work in general for small tumors, and you don't want to miss those. A fixed threshold, perhaps based on some reference if your overall intensity is not stable, would detect tumors of any size.

[Image Analyst]

Go here for tumor finding algorithms: http://www.visionbib.com/bibliography/contentsmedical.html#Medical%20Applications,%20CAT,%20MRI,%20Ultrasound,%20Heart%20Models,%20Brain%20Models