Document Type : Full Research Paper
Authors
1 M.Sc Student, Biomechanic Department, Biomedical Engineering Faculty, Amirkabir University of Technology, Tehran, Iran
2 Ph.D Student, Biomechanic Department, Biomedical EngineeringFaculty, Amirkabir University of Technology, Tehran, Iran
3 Associate Professor, Biomechanic Department, Biomedical EngineeringFaculty, Amirkabir University of Technology, Tehran, Iran
4 Assistant Professor, Functional Neurosurgery Research Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran
Abstract
Spinal deformities are generally associated with lumbar and cervical chronic pain and additionally they disturb the health. In these deformities, lumbar spinal curvature undergone changes in three dimensional space and in most cases, they cause reduction of lung capacities, breathing problems and negative effects on cardiovascular system. In critical deformity cases, in order to correct the deformity and prevent its progression, surgeons determine to perform posterior spinal fusion. As a result, they need to extract some important clinical parameters of spine such as Cobb angle, sagittal and coronal balance, spinal curvature, vertebraes angles and their rotations. In this study, edited tomographic images in MIMICS, were used to prepare a three dimensional model of the spine. Then by using curve fitting techniques and different clustering methods such as self-organization nueral network, k-means and hierarchical method, vertebras were separated and important geometrical data such as curvature of the spine and vertebras angle were obtained. In addition, through implementation of certain algorithms, other clinical features of each vertebra, including minimum and maximum height, length and width of the vertebral body and the relative displacement of vertebras were calculated automatically. In order to validate the proposed methods, measures and angles; derived values obtained automatically at each stage, were again calculated by a radiologist and a spine surgeon who was unaware of the goals of the research. Automatic values were verified by being compared with these manual results. In conclusion the reliability, accuracy and performance of the proposed automatic algorithms were demonstrated.
Keywords
Main Subjects