View Full Version : Medical image smoothing

Barker, Dan [sahc_rgh]
07-21-1998, 06:02 PM
Dear biomechlr's

It is often necessary in biomechanics to determine the distance from a
tendon to the centre or edge of a bone especially for moment arm
calculations. We have acquired MRI transverse sections which provide
sectional views of the tendon and bone.

My question to the group is: Is there any gold standard methodology for
determining what is the edge of bone and tendon in an MRI section?

There are different methods in the literature for example thresholding,
filtering, polar series line fitting. Algorithms such as Roberts and
Sobel can usually be applied but I require one line as a best fit of the
noisy outline with a quantifyable error. This suggests to me that a
polar (ellipse?) fourier series would be best but a threshold value
still has to be determined for the pixels which comprise the 'edge'.

All comments will be greatly appreciated and a summary of replies posted

Dan Barker (sbarkds@rgh.sa.gov.au)
Research Engineer
Repatriation General Hospital
Division of Orthopaedic Surgery
Daws Rd. Daw Park 5041 S.A.
Fax: 61 8 8374 0712
Tel: 61 8 8275 1107

To unsubscribe send UNSUBSCRIBE BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://www.bme.ccf.org/isb/biomch-l