Cyst and tumor in oral cavity are seriously noticed by health experts along with increasing death cases of oral cancer in developing country. Early detection of cyst and tumor using dental panoramic image is needed since its initial growth does not cause any complaints. Image processing is done by mean for distinguishing the classification of cyst and tumor. The results in previous studies about classification of cyst and tumor were done by using a mathematical computation approach namely supports vector machine method that have still not satisfied and have not been validated. Therefore, in this study we propose a method, i.e., nonparametric regression model based on local polynomial estimator that can be improve the classification accuracy of cyst and tumor on human dental panoramic image. By using the proposed method, we get the classification accuracy of cyst and tumor, i.e., 90.91% which is greater than those by using the support vector machine method, i.e., 76.67%. Also, in validation process we obtain that the nonparametric regression model approach gives a significant Press's Q statistical testing value. So, we conclude that the nonparametric regression model approach improves the classification accuracy and gives better outcome to classify cyst and tumor using dental panoramic image than the support vector machine method.
The semiparametric regression is one of the three forms of regression analysis which is made up of parametric and nonparametric. While the parametric is based on linear estimator, this nonparametric component is an innovation. This research proposes all the possible trigonometric basis usually used in Fourier series as nonparametric component estimator, its advantage, which includes its ability to overcome data with oscillation patterns. This study discusses nonparametric regression based on complete and sine Fourier series. Both estimators are developed using the cosine Fourier series concept. The outputs are two estimators which are used for parametric and nonparametric components with the corresponding form in semiparametric regression. In addition, all of these can be applied in real problems, and the best estimator is determined based on the smallest GCV and MSE for an oscillation parameter which gives the highest coefficient of determination for the selected one.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.