Spinal curvature determination from scoliosis X-Ray image using sum of squared difference template matching

Abstract

Scoliosis is a disorder in which there is a sideways curve of the spine. Curve are often S-shaped or C-shaped. One of the methods to ascertain a patient with scoliosis is through using Cobb angle measurement. The importance of the automatic spinal curve detection system is to detect any spinal disorder quicker and faster. This study is intended as a first step based on computer-aided diagnosis. The spinal detection method that we propose is using template matching based on Sum of Squared Difference (SSD). This method is used to estimate the location of the vertebra. By using polynomial curve fitting, a spinal curvature estimation can be done. This paper discusses the performance of SSD method used to detect a variety of data sources of X-Ray from numerous patients. The results from the implementation indicate that the proposed algorithm can be used to detect all the X-ray images. The best result in this experiment has 96.30% accuracy using 9-subdivisions poly 5 algorithm, and the average accuracy is 86.01%.