Dental Morphometric Analysis in Digital Orthopantomogram Using the Software Package - “Shape ver.1.3”

Introduction Advancements in digital tooth recognition and dental morphology analysis using X-ray imaging have revolutionized the diagnostics and treatment planning. For example, new techniques like Convolutional Neural Networks (CNNs) can automate tasks such as detecting and numbering teeth on radiographs [1,2]. Still, it requires a database of digestible information to be trained on. Principal Component Analysis (PCA) is a statistical tool used for dimensionality reduction and pattern recognition. In dental research, PCA can effectively analyze and quantify the shape of teeth variations by breaking down complex structures into principal components (PCs) [3, 4]. Although, the mathematical analysis of PCA may be too complex for medical researchers. That is why the software package “SHAPE ver.1.3” could be a simplistic way to create a digital dental database, which could be used to advance AI for developing a teeth recognition system. Aim The aim of the present study is to determine the suitability of the PCA in dental morphometrics to digitize orthopantomograms (OPG), using program package "SHAPE ver.1.3", and the ability to compare digitized teeth morphology. Methods An OPG of a patient, who had all 32 of their teeth erupted and presented with a physiological x-ray image, was chosen to be evaluated. Firstly, all the contours of the 32 teeth were traced using an image editing software “GIMP 2.10.38”, fitted into a correct format and digitized using "SHAPE ver.1.3" software’s programs: “ChainCoder“, “Chc2Nef“, “PrinComp”, “PrinPrint”. Principal component analysis was carried: the number of PCs, their eigenvalues and proportion of each PC were calculated for each of the teeth. Results Every single tooth can be evaluated using 4 principal components. The tooth with the highest proportion of the first PC was 36 (PC1 = 1.942838E-002, proportion = 97.584%, PC2 = 2.944219E-004, proportion = 1.4788%, PC3 = 1.324422E-004, proportion = 0.6652%, PC4 = 5.414024E-005, proportion = 0.2719%). The tooth with the lowest proportion of the first PC was 31 (PC1 = 6.535262E-005, proportion = 36.6719 %, PC2 = 4.928452E-005, proportion = 27.6554 %, PC3 = 4.224883E-005, proportion = 23.7074 %, PC4 = 2.132319E-005, proportion = 11.9653 %). Overall, the posterior teeth (4-8) showed a higher proportion of PC1 (mean - 68.9, std. deviation - 17.63) than front teeth (1-3) (mean - 53.4, std. deviation - 13.44) in all quadrants. Conclusions Our results indicate that since every tooth in OPG consists of four PCs, it is possible to evaluate and digitize teeth shapes in orthopantomograms using PCA with the ,,Shape ver. 1.3” software package. Moreover, findings from digitized images reveal that the molars and premolars have a more mathematically correct anatomy, thus comparing anatomy is possible.