A computerized method for evaluating scoliotic deformities using elliptical pattern recognition in X-ray spine images

Area of Science:

• Medical Imaging
• Biomechanical Engineering
• Computational Geometry

Background:

• The Cobb angle is a standard metric for assessing scoliosis severity from X-ray images.
• Previous research indicates ellipses can model spinal curvature but haven't been used for scoliosis quantification.
• Existing methods for measuring spinal deformities have limitations in accuracy and observer interaction.

Purpose of the Study:

• To develop and validate a novel computerized methodology for detecting elliptical patterns in X-ray spine images.
• To quantify scoliotic deformities using the geometric parameters of best-fit ellipses.
• To establish a new index for evaluating spinal curvature in scoliosis patients.

Main Methods:

• Spine curves were reconstructed from vertebral centroids in anteroposterior X-ray images.
• A least squares and genetic algorithm optimization framework was employed to find the best-fit ellipse.
• Geometric parameters of the ellipse were used to define a spinal curvature index.

Main Results:

• The methodology achieved a maximal relative error of 3% on synthetic images.
• Validation on 20 clinical X-ray images showed an overall error of 0.5 ± 0.4 mm.
• The method reliably reproduced scoliotic curvatures using ellipse parameters.

Conclusions:

• The proposed computerized method accurately quantifies scoliosis using elliptical pattern detection.
• This approach offers potential advantages over conventional methods, including reduced error and observer interaction.
• The methodology provides a comprehensive tool for scoliosis diagnosis, study, and treatment planning.