The anatomies of pelvic structures are critical for the diagnosis of pelvic floor dysfunctions. However, because of the complex background, the imaging appearances of pelvic organs and muscles are frequently distorted by noise and partial volume effect. Magnetic resonance imaging with its clear imaging quality of the female pelvic cavity is preferred for many studies. As such, correct segmentations of the pelvic structures on MR images are required for accurate diagnoses. Effective algorithms for axial T2-weighted MR images have been proposed, which are based on the imaging features of different structures and various image clues. In this paper, we review these algorithms and evaluate their performance, and discuss implementation issues and aspects towards constructing the three-dimensional models.
Keywords: bladder; deformable models; image segmentation; levator ani muscle; medical imaging; rectum; vagina.
Copyright © 2012 John Wiley & Sons, Ltd.