Predictor Identification of Successful Treatment in Class III Malocclusion Using Feature Selection

摘要

Class III malocclusion where the lower teeth overlaps the upper teeth is common dental problem among Asians. Many researchers have attempted to predict the prognosis of early treatment in Class III malocclusion successfully by identifying the dental markers that can efficiently discriminate patient samples with good and poor prognoses. However, since these approaches do not consider classification accuracy in examining the discrimination power of a marker to classify patient samples, the identified markers may not achieve the highest accuracy in the prognosis prediction. In this paper, therefore, we proposed a general framework using a feature selection method for dental marker identification and optimal classifier generation showing the highest classification accuracy of prognosis prediction in Class III malocclusion. From a series of experiments, our results show that the dental markers identified by feature selection method can attain very high classification accuracy. Although, in this work, we focus on the marker identification of Class III malocclusion, the proposed scheme is expected to deal with more broad range of dental problems such as Class I and II malocclusions.