An active shape model for the reconstruction of scoliotic deformities from back shape data |
| |
Authors: | Huysmans T Moens P Van Audekercke R |
| |
Affiliation: | Department of Mechanical Engineering, Division of Biomechanics and Engineering Design, Katholieke Universiteit Leuven, Celestijnenlaan 200 A, 3001 Heverlee, Belgium. tom.huysmans@mech.kuleuven.ac.be |
| |
Abstract: | OBJECTIVE. The objective of the present study is to improve the accuracy and reliability of the spinal midline reconstructions in scoliosis from back shape data. Design. An active shape model which covers the variety of scoliotic curves with a minimum of adjustable parameters is designed for the reconstruction of the spinal midline based on rasterstereographic back surface measurements. BACKGROUND. To reduce the number of X-rays needed for patients with a scoliosis, an automatic method was developed for the reconstruction of spinal deformities from back shape data. METHODS. 264 digital X-rays of 264 patients with scoliosis were used as a training set. By examining the statistics of the vertebral body centres a point distribution model was derived. The model is used as a basis to reconstruct the spinal midline from back shape data (active shape model). RESULTS. 478 rasterstereographs of 114 scoliotic patients have been evaluated with this new procedure and with an existing procedure. Both procedures deliver a three-dimensional curve of the spinal midline. The frontal projections of these spinal midlines are compared with the vertebral centres of the corresponding 478 X-rays. The active shape model improved the results as compared to the existing procedure from sigma(x)=3.4mm to sigma(x)=3.0mm. CONCLUSION. The use of the active shape model improves the overall accuracy of the spinal midline reconstructions in scoliosis from back shape data.RELEVANCE. Improvements to the accuracy and the reliability of the three-dimensional spinal reconstruction based on back shape data, can lead to an additional reduction of X-rays for patients with a scoliosis. |
| |
Keywords: | |
本文献已被 PubMed 等数据库收录! |
|