CAMS: A Cobb Angle Measurement Software via Extracting Spinal Curvature in X-ray Images

Scoliosis is a musculoskeletal disorder in which a person's spine bends sideways and rotates along its vertical axis. Cobb-angle (CA) is a common index to evaluate, determine, and track the progression of scoliosis; however, measuring this index manually depends on the operator’s experience and image quality, which may lead to errors, dispersion of obtained values, and variability. This study aimed to resolve this issue by developing an algorithm for measuring CA.

In this analytical study, we developed software using image processing, data modeling, and analytical geometry tools for measuring CA.  A given x-ray image was processed to highlight the spinal cord. Then, the spinal cord curvature was extracted by manually segmenting the spinal cord in the given image and fitting a polynomial function to the data points in the identified region.  Finally, CA was estimated by calculating the angle between the two normal lines that pass through the inflection points of the fitted curve. Thirty X-ray images having CA values obtained by an expert were used to evaluate the accuracy of the developed algorithm.

There is no statistically significant difference between the average CA values measured by the expert (53.95º ± 20.3º) and those estimated using the developed software (56.41º ± 19.95º). There is a significant correlation between the values estimated using the developed algorithm and the reference values, (r=0.93) and (P<10-8).

The obtained results are promising and show that the developed algorithm might be used to measure cobb-angle. Nevertheless, its accuracy and reliability should be further evaluated using a large data set.