This paper presents a comparative study of machine learning algorithms for classifying normal, abscessed, and impacted tooth based on periapical radiograph images. Those methods are Logistic Regression (LR), Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), Random Forest (RF), Gaussian Naive Bayes (NB), and Support Vector Machine (SVM). Haralick texture, Hu’s moment invariants, and color histogram are utilized to obtain the feature vector of those images. The accuracy can be calculated with 10-fold cross-validation. We also verify the accuracy of the machine learning algorithms under the various number of training images. We take 30, 45, and 60 images from three classes. Regardless the number of training images, RF keeps outperforming the others in the term of accuracy.
https://ieeexplore.ieee.org/document/8534887
- Published in: 2018 2nd International Conference on Biomedical Engineering (IBIOMED)
- Date of Conference: 24-26 July 2018
- Date Added to IEEE Xplore: 15 November 2018
- ISBN Information:
- Electronic ISBN: 978-1-5386-4736-3
- Print on Demand(PoD) ISBN: 978-1-5386-4737-0
- DOI: 10.1109/IBIOMED.2018.8534887
- Publisher: IEEE
- Conference Location: Bali, Indonesia, Indonesia
Keywords:
IEEE Keywords:
Teeth, Dentistry, Machine learning algorithms, Radio frequency, Support vector machines, Image color analysis, Forestry.
Author Keywords:
Periapical Radiograph, Logistic Regression, Linear Discriminant Analysis, K-Nearest Neighbors, Random Forest, Gaussian Naive Bayes, Support Vector Machine.
Conference call IEEE.