0

I have several x-ray images of the same bone taken at different times (during followups), approximately from the same angle of the camera plane compared with the bone surface plane. However, the angle, location and size of the bone in the images are different.

My goal is to align the images using the best registration technique for such images (automatically, that is, without having to manually pick out registration points), using MATLAB.

I have tried using this code from mathworks.com : http://www.mathworks.com/help/images/examples/find-image-rotation-and-scale-using-automated-feature-matching.html and it worked perfectly for recognizing points on the implant and aligning the images according to those points, but I cannot seem to get it to recognize similar bone patterns.

To help "focus" on the bones instead of the implant, I have removed the implant completely from the image. Still, detectSURFFeatures caught no bone features to work with. Changing the parameters of detectSURFFeatures didn't help much either.

I wish to stick to this overall code because it does work very well as long as I can get it to detect enough feature points.

Does anyone have any suggestions on how to detect bone features/patterns such as in joints or fractures (using detectSURFFeatures or any other similar function)?

Thank you, Noga

Noga
  • 41
  • 2
  • 3
    Can you provide a link with sample images ? – Ratbert Jul 20 '15 at 09:54
  • I wasn't able to upload the images, perhaps because they are very large. However I was able to find online this fairly similar example of two x-ray images requiring alignment: http://www.radiologyassistant.nl/data/bin/w440/a5097977a171e7_stress1.jpg – Noga Jul 20 '15 at 14:04

1 Answers1

2

Check out the work by Modersitzki:

FAIR: Flexible Algorithms for Image Registration - Software and Apps

It is primarily focused on registration of medical images.

There is also a MATLAB toolbox with examples available for download.