Clustering Validation of Spinal Deformity Classification With Simplified 3-D Model and Principal Component Analysis

In this paper, the clustering validation of spinal deformity classification by principal component analysis is introduced for the visualization of high dimensional patterns of the scoliosis spinal deformity with their reduced two-dimensional clustering properties. King spinal deformity classification system was used for PCA cluster implementation. The dataset used for this study had 25 spinal deformity patterns. Such dataset was further divided into 5 subgroups. In each subgroup there were exactly 5 King deformity patterns. These 25 King spinal deformity classifications had been verified by artificial neural network with hold-out method. At the first step the simplified three-dimensional (3-D) spine model was constructed based on the coronal and sagittal X-ray images. The features of the central axis curve of the spinal deformity patterns in 3-D space were extracted by the total curvature analysis. The discrete form of the total curvature, including the curvature and the torsion of the central axis of the simplified 3-D spine model was derived from the difference quotients. The total curvature values of 17 vertebrae from the first thoracic to the fifth lumbar spine formed a Euclidean space of 17 dimensions. For the purpose of classification, each pattern of spinal deformity was labeled to identify it from others. Principal component analysis (PCA) was applied on the total curvature of these 25 spinal deformity patterns. By PCA dimensional reduction on these patterns, the 17-D pattern characters of spinal deformity were reduced into 2-D and the clustering property of spinal deformity classification can be visualized in a 2-D principal component plane.