Determine the area of the nodules (in pixels)

I'm currently working on a school project of medical imaging. I have an X-Ray image of hands with nodules. The objective is to determine the area of the nodules (in pixels). To achieve the requested objective, we must work on the image, apply some of the methods taught, improve it, and subsequently segment structures of interest.

How can i do it?

Thank you everyone in advance.

[Cris LaPierre]

Segment your image, then use regionprops to compute the area in pixels of each label.

[Joana]

How can i do the segmentation of this image?

[Walter Roberson]

I am not at all certain that it is possible to reliably differentiate between fingers and nodules.

[Cris LaPierre]

You said you need to apply the methods you were taught. What methods are you expected to use?

[Joana]

we must work on the image, apply some of the methods taught (such as change the brightness and contrast of the image, binarization, segmentation) improve it, and subsequently segment the structures of interest, which are the nodules on the fingers

Answer 1

[Image Analyst]

It looks rather tough since the abnormalities are of about the same brightness as the bones. But there does seem to be a difference in texture. So I'd try a texture filter such as stdfilt or entropyfilt. Try different window sizes and see how the image changes.

You can further segment blobs with size filters like bwareafilt or bwareaopen. I seriously doubt global brightness or contrast adjustment will help any. Just because you make the image easier to see does not mean the binarization/thresholding will improve - it will merely happen at a different threshold value afterwards than before adjustment/enhancement.

I also have doubts about local contrast enhancement like adapthisteq will help but you can try it.