Abstract

Background

The aim was to develop statistical shape models of the main human tarsal bones that would result in novel representations of cuboid, navicular and talus.

Methods

Fifteen right and 15 left retrospectively collected computed tomography data sets from male individuals, aged from 17 to 63 years, with no known foot pathology were collected. Data were gathered from 30 different subjects. A process of model building includes image segmentation, unifying feature position, mathematical shape description and obtaining statistical shape geometry.

Results

Orthogonal decomposition of bone shapes utilising spherical harmonics was employed providing means for unique parametric representation of each bone. Cross-validated classification results based on parametric spherical harmonics representation showed high sensitivity and high specificity greater than 0.98 for all considered bones.

Conclusions

The statistical shape models of cuboid, navicular and talus created in this work correspond to anatomically accurate atlases that have not been previously considered. The study indicates high clinical potential of statistical shape modelling in the characterisation of tarsal bones. Those novel models can be applied in medical image analysis, orthopaedics and biomechanics in order to provide support for preoperative planning, better diagnosis or implant design.

Details

Title
Statistical shape models of cuboid, navicular and talus bones
Author
Melinska, Aleksandra U; Romaszkiewicz, Patryk; Antosik, Justyna Wagelrtlomiej; Sasiadek, Marek; Iskander, D Robert
Publication year
2017
Publication date
2017
Publisher
BioMed Central
e-ISSN
1757-1146
Source type
Scholarly Journal
Language of publication
English
ProQuest document ID
1865001515
Copyright
Copyright BioMed Central 2017