How can I find inner and outer boundaries of binary image

Hi,

I am working on a segmentation project, where I could successfully segment the questionable area and I got a binary image. I used bwboundaries() in order to get the boundaries, nevertheless it found one continuoes boundary instead of two ( inner and outer).

Are there any way to detect the inner- and outer boundaries separately, like this?

Thanks for your suggestions!

[Matt J]

What defines the beginning/ending points of the red and blue boundaries?

[Máté Kiss]

The corner points of the segmented volume.

[Image Analyst]

For some arbitrary blob shape, like a kidney bean, or an asterisk, or whatever, you have no way to define the splitting point of the boundary. Just saying "the corner points" is not good enough. You're going to have to define it better than that. There could be lots of places where the change in direction is high. Try to explain in more detail what it is about those points you chose. And what do your other masks look like? Are they all U or V-shaped blobs? What about taking the convex hull and then finding the largest interior bay and finding where that boundary intersects your outer hull?

Answer 1

[Máté Kiss]

Well, here is the original image, after cropping (cardiac MR scan). I also attached a mat file.

Then I applied otsu segmentation to differentiate the background, then bwboundaries. I've tried convhull, but e.g. in that case it can computes only one convhull - it doesn't help a lot.

regarding to the starting- and endpoint definition, I attached an image about the expected start and endpoint. If I would close the open part of the heart (see the image), the outer boundary's starting point would be the closest point to the closing line (1) to end point would be the second closest point to the closing line (2). It doesn't matter which one is which.