Determination of spinal curvature from scoliosis X-ray images using K-means and curve fitting for early detection of scoliosis disease

One of the disease that require X-ray diagnosis is scoliosis. Early detection of scoliosis is important to do for anyone. From the early detection information, the doctor may take the firts step to further treatment quickly. Determination of spinal curvature is a first step method that used to measure how severe the degree of scoliosis. The severity degree of scoliosis can be assess by using Cobb angle. Therefore, by approximate the spinal curvature, we can approximate the cobb angle too. From previous work that interobserver measurement value may reach 11.8° and intraobserver measurement error is 6°. So, as far as the cobb angle measuring, the subjectivity aspect is the natural thing and can be tolerated until now. This research propose an algorithm how to define spinal curvature with the aid of a computer in digital X-ray image quickly but has a standard error that can be tolerated. The preprocessing has been done by canny edge detection. The k-means clustering algorithm can detect the centroid point after segmentation preprocessing of the spinal segment and polynomial curve fitting will be used in the process for determining the spinal curve. From the spinal curvature information, the scoliosis curve can be classified into 4 condition, normal, mild, moderate, and severe scoliosis.