Artificial Neural Networks Can be Used as Alternative Method to Estimate Loss Tooth Root Sizes for Prediction of Dental Implants

Abstract

The aim of this study was to investigate the feasibility of estimation of canine root length and cervical width by an artificial neural network method with an appropriate setting. We randomly obtained 120 representative samples of routine panoramic radiographs by computer tomography (CT). Of 120 samples, 96 (80%) were used in training phase and 24 (20%) were used in test phase after a randomized selection. The intertuberal length (IL) of maxilla and canine root length (RL) and canine root cervical width (CW) of the right canine tooth was measured and was entered to a datum file. According to the results, the method is convenient with this purpose. The mean square error values are lied between 2% and 4.4% for the estimations. This shows that ANN is an alternative method for the prediction of canine root length and cervical width. Since an ANN software, as a cost-effective tool that can be purchased and after a training procedure that can be easily performed by an experienced person, the ANN system can have a place in implant surgery after further studies and optimization of software setting. An ANN software can be adapted to the dental implant surgery after proper setting and training procedures that can be easily performed by an experienced person and it can have a place in implant surgery after further expanded studies and optimization of software setting.