Principal Component Analysis for Clustering Temporomandibular Joint Data

Temporomandibular joint disorders are the second most common cause of orofacial pain. The replacement of symptomatic, degenerative temporomandibular joint is attracting increasing attention. It is important that any surgical implants has the correct dimensions to fit its intended site. Cluster analysis is a method to determine the size ranges for temporomandibular joint replacement prosthesis. In this study, the knowledge-based K-means clustering method was proposed. Principal component analysis is a classical technique to reduce the dimensionality of the data set by transforming to a new set of variables to summarize the features of the data. In our paper, we study the effectiveness of principal components in capturing cluster structure. Specifically, we compare the quality of clusters obtained from the original data to the quality of clusters obtained after projecting onto subsets of the principal component axes. Our study shows that clustering with the principal component instead of the original variables will improve the cluster quality. Overall, we would recommend principal component analysis before clustering in Euclidean space.