Presentation Category
Research
Introduction/Context/Diagnosis
Abstract: Objectives: To compare facial growth prediction models based on the partial least squares and artificial intelligence (AI) Materials and Methods: Serial longitudinal lateral cephalograms from 33 patients who had not undergone orthodontic treatment but had taken serial cephalograms were collected from Mathews Growth Study located in the American Association of Orthodontists Foundation (AAOF) Growth Legacy Collection. On every image, 46 skeletal and 32 soft- tissue landmarks were identified using CEPPRO. Growth prediction models were built using multivariate Partial Least Squares regression (PLS) and a deep learning method based on the deep neural network incorporating 161 predictors, and 156 responses, variables. The prediction accuracy between the two methods was compared. Results: On average, AI showed less prediction error than PLS. Among the 78 landmarks, AI was more accurate in 36 landmarks, whereas PLS was more accurate in 6 landmarks. The remaining 36 landmarks showed no statistical difference between the two methods. Overall, soft-tissue landmarks and landmarks in the mandible showed greater prediction errors than hard-tissue landmarks and landmarks in the maxilla, respectively. Conclusions: PLS and AI methods seemed to be valuable tools for predicting growth. PLS accurately predicted landmarks with low variability in the cranial base. In general, however, AI outperformed, particularly for those landmarks in the maxilla and mandible. Applying AI for growth prediction might be more advantageous when uncertainty is considerable. Acknowledgement: We appreciate Dr. Heesoo Oh and Dr. Heeyeon Suh for guiding and helping us with the research and poster. Thanks to your advice and mentorship, we could learn how to analyze and interpret the data and how to apply the results of comparison between AI and PLS when predicting growth for real patients.
Location
Arthur A Dugoni School of Dentistry, 155 5th St, San Francisco, CA 94103, USA
Format
Presentation
Growth Prediction Using Artificial Intelligence
Arthur A Dugoni School of Dentistry, 155 5th St, San Francisco, CA 94103, USA
Abstract: Objectives: To compare facial growth prediction models based on the partial least squares and artificial intelligence (AI) Materials and Methods: Serial longitudinal lateral cephalograms from 33 patients who had not undergone orthodontic treatment but had taken serial cephalograms were collected from Mathews Growth Study located in the American Association of Orthodontists Foundation (AAOF) Growth Legacy Collection. On every image, 46 skeletal and 32 soft- tissue landmarks were identified using CEPPRO. Growth prediction models were built using multivariate Partial Least Squares regression (PLS) and a deep learning method based on the deep neural network incorporating 161 predictors, and 156 responses, variables. The prediction accuracy between the two methods was compared. Results: On average, AI showed less prediction error than PLS. Among the 78 landmarks, AI was more accurate in 36 landmarks, whereas PLS was more accurate in 6 landmarks. The remaining 36 landmarks showed no statistical difference between the two methods. Overall, soft-tissue landmarks and landmarks in the mandible showed greater prediction errors than hard-tissue landmarks and landmarks in the maxilla, respectively. Conclusions: PLS and AI methods seemed to be valuable tools for predicting growth. PLS accurately predicted landmarks with low variability in the cranial base. In general, however, AI outperformed, particularly for those landmarks in the maxilla and mandible. Applying AI for growth prediction might be more advantageous when uncertainty is considerable. Acknowledgement: We appreciate Dr. Heesoo Oh and Dr. Heeyeon Suh for guiding and helping us with the research and poster. Thanks to your advice and mentorship, we could learn how to analyze and interpret the data and how to apply the results of comparison between AI and PLS when predicting growth for real patients.
Comments/Acknowledgements
Presentation Category: Research