Physical-based statistical shape modeling of the levator ani

IEEE Trans Med Imaging. 2009 Jun;28(6):926-36. doi: 10.1109/TMI.2009.2012894. Epub 2009 Jan 19.

Abstract

The levator ani is the main muscular support of the pelvic floor organs and damage caused by childbirth can affect its function. The full functionality of this muscle group is still unknown but it is essential for effective surgical planning. To elucidate its functional significance, a physical-based statistical shape model was built from the levator ani surfaces of 15 subjects scanned in an open access scanner. Simulation of dynamic exercises was performed on the resulting surfaces with finite element analysis. Statistical shape modeling was performed on the training set consisting of the original and simulated shapes along with thickness and strain distributions. Simulation results are presented on 15 subjects. The statistical shape model shows good correspondence to inter- and intra-subject shape variability, with the modes of variation highlighting movement in the posterior of the levator ani as well as in the levator arms. Strain distribution plots and the modes of variation show results that correspond to clinical findings. Further validation of the technique and a repeatability test were performed on four subjects with internal global pressure readings taken from a perineometer and five patients suffering from minor pelvic floor disorders due to obstructed defaecation. A Mann-Whitney nonparametric test was used to compare the normal model fitting to the two subject groups.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Algorithms
  • Biomechanical Phenomena
  • Computer Simulation
  • Elasticity
  • Female
  • Finite Element Analysis
  • Humans
  • Magnetic Resonance Imaging
  • Models, Anatomic
  • Models, Statistical*
  • Muscle Contraction*
  • Pelvic Floor / anatomy & histology*
  • Pelvic Floor / physiology*
  • Perineum / physiology
  • Principal Component Analysis
  • Reproducibility of Results
  • Statistics, Nonparametric