How to use machine learning in image segmentation?

I have an image database containing raw medical images (lung x-ray image) and their corresponding binary masks indicating blood vessels. I would like to apply machine learning techniques suck as GLM on these training data to build a model. So I can use this model to generate binary blood vessle mask for other lung x-ray images. I need some help for a machine learning workflow/pseudo code for this project.

The Matlab code

http://www.mathworks.com/matlabcentral/fileexchange/46082-retinal-vessel-detection-by-gabor-transform-and-machine-learning--a-tutorial

is a good start but I'm not sure how I can apply this code to a entire image library?

Answer 1

[Image Analyst]

Assuming your "library" is a collection of individual image files, you can follow the FAQ:

http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

In the middle of the for loop, you could call AnalyzeSingleImage() which is a function you write that has all of that MATLAB code in it that you got (and probably modified) from the File Exchange.

[Kyle]

not really right. I know how to process a sequence of images, but my concerns is using machine learning to learn the correlation between the raw image and binary mask.

[Image Analyst]

Well, I was answering your last question where you asked "how I can apply this code to a entire image library?" You asking that question led me to believe this was not the case: "I know how to process a sequence of images"

Sorry, but we have not run that File Exchange submission, so if you have a question about that, your best (only) approach is to ask the author who wrote and uploaded it. No one else will know about it.

[Salma Hassan]

can i use the weka application for this problem . the input is the attributes like (glcm features ) and choose the classifier from weka