Enhancing Bioprogressive Dentofacial-Skeletal Complex Growth Prediction Utilizing Digital Datamining

Objectives: To evaluate and compare the accuracy of two computerized growth prediction methods based on lateral cephalographs utilizing the Bolton-Broadbent and Ricketts normative growth data algorithm as embedded in the Dolphin Imaging diagnostic computer system.

Methods: Longitudinal sets of non-treated (control) cephalometric radiographs from 36 patient records (20 males and 16 females) were acquired from the collection of Dr. Robert Ricketts.147 different growth predictions were performed (59, 46 and 42 combinations derived for 2 years, 3 years, and 4 years prediction respectively from 93 males and 54 females combinations). The subjects' lateral cephalometric films were scanned for each time-point and traced using commercially-available imaging software. Subjects and their associated radiograph series were grouped by gender and age. Each algorithm was used to run 2, 3 and 4 years of growth prediction. The result of each prediction tracing was superimposed upon the corresponding tracing of the same control subject within the same age group of prediction year. Superimpositions were done using structural methods and best fit to visualize growth and craniofacial development. Linear and angular differences in hard and soft tissues cephalometric landmarks were calculated. One sample t-tests and pairwise student t-tests for each landmark were done (p=0.05) to assess statistical differences. Accuracy of the prediction algorithm were tested for 62 variables (8 cephalometric value changes and 54 linear distance changes).

Results: The Bolton-Broadbent algorithm prediction was between 73-84% accurate in predicting the X & Y plane changes for males; and 45-53% accurate for the female groups. The Ricketts algorithm was not more than 50% accurate in its predictions for either group.

Conclusions: Both algorithms failed to accurately predict the cephalometric measurements. For both algorithms, there was no statistical difference between the 2, 3, or 4 year groups. There was no statistical difference between the male or female groups.