Orthodontic Treatment Planning based on Artificial Neural Networks

Peilin Li, Deyu Kong, Tian Tang, Di Su, Pu Yang, Huixia Wang, Zhihe Zhao, Yang Liu, Peilin Li, Deyu Kong, Tian Tang, Di Su, Pu Yang, Huixia Wang, Zhihe Zhao, Yang Liu

Abstract

In this study, multilayer perceptron artificial neural networks are used to predict orthodontic treatment plans, including the determination of extraction-nonextraction, extraction patterns, and anchorage patterns. The neural network can output the feasibilities of several applicable treatment plans, offering orthodontists flexibility in making decisions. The neural network models show an accuracy of 94.0% for extraction-nonextraction prediction, with an area under the curve (AUC) of 0.982, a sensitivity of 94.6%, and a specificity of 93.8%. The accuracies of the extraction patterns and anchorage patterns are 84.2% and 92.8%, respectively. The most important features for prediction of the neural networks are "crowding, upper arch" "ANB" and "curve of Spee". For handling discrete input features with missing data, the average value method has a better complement performance than the k-nearest neighbors (k-NN) method; for handling continuous features with missing data, k-NN performs better than the other methods most of the time. These results indicate that the proposed method based on artificial neural networks can provide good guidance for orthodontic treatment planning for less-experienced orthodontists.

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
(a) The data processing flow chart; (b) structure of the neural network to predict the extraction patterns. The network structure is a three-layer fully connected multilayer perceptron consisting of 24 input nodes, 10 hidden nodes, and 4 output nodes.
Figure 2
Figure 2
The ROC curve of the neural network to predict extraction. The model yields an AUC of 0.982 (95% CI 0.968–0.995). The optimum diagnostic cutoff value is 0.692, where the sensitivity of the model reaches 94.6% (95% CI 0.894–0.964) and the specificity reaches 93.8% (95% CI 0.870–0.984).
Figure 3
Figure 3
The accuracies of the ANNs. The accuracy of the extraction-nonextraction prediction is 94.0%, and the accuracies of the learning set, the validation set, and the test set are 94.0%, 95.0% and 93.3%, respectively. The predictive accuracy of the extraction patterns is 83.3%, and the accuracies of the learning set, validation set, and test set are 83.6%, 84.1%, and 81.8%, respectively. The overall accuracy of the anchorage patterns is 92.8%, and the accuracies of the learning set, validation set, and test set are 93.3%, 90.9%, and 93.2%, respectively.
Figure 4
Figure 4
Clinical application illustration of the ANNs. The medical records of a new case were collected, and 24 input features, including demographic data, cephalometric data, dental data and soft tissue data, were extracted for neural network prediction. The extraction probability (0.955) was higher than 0.692; thus, it was determined as an extraction case and was passed to the other two networks. The other networks output the feasibilities of different extraction patterns and anchorage patterns. The doctor evaluated these treatment options, took other aspects into account, and finally determined an effective treatment plan.

References

    1. Krooks L, Pirttiniemi P, Kanavakis G, Lahdesmaki R. Prevalence of malocclusion traits and orthodontic treatment in a Finnish adult population. Acta odontologica Scandinavica. 2016;74:362–367. doi: 10.3109/00016357.2016.1151547.
    1. Feldens, C. A. et al. Impact of malocclusion and dentofacial anomalies on the prevalence and severity of dental caries among adolescents. The Angle orthodontist, 10.2319/100914.1 (2015).
    1. Mary AV, et al. Assessing Quality of Life using the Oral Health Impact Profile (OHIP-14) in Subjects with and without Orthodontic Treatment need in Chennai, Tamil Nadu, India. Journal of clinical and diagnostic research: JCDR. 2017;11:Zc78–zc81. doi: 10.7860/jcdr/2017/27309.10442.
    1. Proffit WR, Fields HW, Jr., Moray LJ. Prevalence of malocclusion and orthodontic treatment need in the United States: estimates from the NHANES III survey. The International journal of adult orthodontics and orthognathic surgery. 1998;13:97–106.
    1. Association, A. D. Oral Health and Well-Being in the United States, (2015).
    1. Proffit, W. R., Fields, H. W. Jr. & Sarver, D. M. Contemporary orthodontics. 5th edn, (Mosby, 2013).
    1. Hammond R, Freer TJ. Application of a case-based expert system to orthodontic diagnosis and treatment planning: a review of the literature. Australian orthodontic journal. 1996;14:150.
    1. Noroozi H. Orthodontic treatment planning software. American journal of orthodontics and dentofacial orthopedics. 2006;129:834–837. doi: 10.1016/j.ajodo.2006.02.025.
    1. Takada K, Yagi M, Horiguchi E. Computational formulation of orthodontic tooth-extraction decisions. Part I: to extract or not to extract. The Angle orthodontist. 2009;79:885–891. doi: 10.2319/081908-436.1.
    1. Yagi M, Ohno H, Takada K. Computational formulation of orthodontic tooth-extraction decisions. Part II: which tooth should be extracted? The Angle orthodontist. 2009;79:892–898. doi: 10.2319/081908-439.1.
    1. Worden K, Staszewski WJ, Hensman JJ. Natural computing for mechanical systems research: A tutorial overview. Mechanical Systems and Signal Processing. 2011;25:4–111. doi: 10.1016/j.ymssp.2010.07.013.
    1. Jia F, Lei Y, Lin J, Zhou X, Lu N. Deep neural networks: A promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data. Mechanical Systems and Signal Processing. 2016;72:303–315. doi: 10.1016/j.ymssp.2015.10.025.
    1. Xie X, Wang L, Wang A. Artificial neural network modeling for deciding if extractions are necessary prior to orthodontic treatment. The Angle orthodontist. 2010;80:262–266. doi: 10.2319/111608-588.1.
    1. Jung SK, Kim TW. New approach for the diagnosis of extractions with neural network machine learning. American journal of orthodontics and dentofacial orthopedics: official publication of the American Association of Orthodontists, its constituent societies, and the American Board of Orthodontics. 2016;149:127–133. doi: 10.1016/j.ajodo.2015.07.030.
    1. Lee R, Macfarlane T, O’Brien K. Consistency of orthodontic treatment planning decisions. Clinical Orthodontics & Research. 1999;2:79–84. doi: 10.1111/ocr.1999.2.2.79.
    1. Luke LS, Atchison KA, White SC. Consistency of patient classification in orthodontic diagnosis and treatment planning. Angle Orthodontist. 1998;68:513.
    1. Sterne JA, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. British medical journal. 2009;339:157–160.
    1. Jerez JM, et al. Missing data imputation using statistical and machine learning methods in a real breast cancer problem. Artificial intelligence in medicine. 2010;50:105–115. doi: 10.1016/j.artmed.2010.05.002.
    1. García-Laencina PJ, Abreu PH, Abreu MH, Afonoso N. Missing data imputation on the 5-year survival prediction of breast cancer patients with unknown discrete values. Computers in biology and medicine. 2015;59:125–133. doi: 10.1016/j.compbiomed.2015.02.006.
    1. Aydilek IB, Arslan A. A novel hybrid approach to estimating missing values in databases using k-nearest neighbors and neural networks. International Journal of Innovative Computing, Information and Control. 2012;7:4705–4717.
    1. Jayalakshmi, T. & Santhakumaran, A. In Data Storage and Data Engineering (DSDE), 2010 International Conference on. 159–163 (IEEE, 2010).
    1. García-Laencina PJ, Sancho-Gómez J-L, Figueiras-Vidal AR. Pattern classification with missing data: a review. Neural Computing and Applications. 2010;19:263–282. doi: 10.1007/s00521-009-0295-6.
    1. Ennett CM, Frize M, Walker CR. Influence of missing values on artificial neural network performance. Stud Health Technol Inform. 2001;84:449–453.
    1. Chen H, Grant-Muller S, Mussone L, Montgomery F. A study of hybrid neural network approaches and the effects of missing data on traffic forecasting. Neural Computing & Applications. 2001;10:277–286. doi: 10.1007/s521-001-8054-3.
    1. Gevrey M, Dimopoulos I, Lek S. Review and comparison of methods to study the contribution of variables in artificial neural network models. Ecological modelling. 2003;160:249–264. doi: 10.1016/S0304-3800(02)00257-0.
    1. Batista GE, Monard MC. A Study of K-Nearest Neighbour as an Imputation Method. HIS. 2002;87:251–260.
    1. Martina, R., Teti, R., D’Addona, D. & Iodice, G. Neural network based system for decision making support in orthodontic extractions. Intelligent Production Machines and Systems, 235–240 (2006).
    1. Dedecker AP, et al. Application of artificial neural network models to analyse the relationships between Gammarus pulex L. (Crustacea, Amphipoda) and river characteristics. Environmental monitoring and assessment. 2005;111:223–241. doi: 10.1007/s10661-005-8221-6.
    1. Graber, L. W., Vanarsdall, R. L., Vig, K. W. & Huang, G. J. Orthodontics: Current Principles and Techniques. (Elsevier Health Sciences, 2016).
    1. Guyon, I. A scaling law for the validation-set training-set size ratio. AT&T Bell Laboratories, 1–11 (1997).
    1. Mavani, V., Raman, S. & Miyapuram, K. P. In Computer Vision Workshop (ICCVW), 2017 IEEE International Conference on. 2783–2788 (IEEE).
    1. Yang, J. B., Nguyen, M. N., San, P. P., Li, X. L. & Krishnaswamy, S. In 2015 International Conference onArtificial Intelligence (Ijcai). 3995–4001.
    1. Kshirsagar A, Seftel A, Ross L, Mohamed M, Niederberger C. Predicting hypogonadism in men based upon age, presence of erectile dysfunction, and depression. International journal of impotence research. 2006;18:47. doi: 10.1038/sj.ijir.3901369.
    1. Boixo S, et al. Characterizing quantum supremacy in near-term devices. Nature Physics. 2018;14:595. doi: 10.1038/s41567-018-0124-x.
    1. Møller MF. A scaled conjugate gradient algorithm for fast supervised learning. Neural networks. 1993;6:525–533. doi: 10.1016/S0893-6080(05)80056-5.
    1. Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout: a simple way to prevent neural networks from overfitting. The Journal of Machine Learning Research. 2014;15:1929–1958.
    1. Hinton, G. E., Srivastava, N., Krizhevsky, A., Sutskever, I. & Salakhutdinov, R. R. Improving neural networks by preventing co-adaptation of feature detectors. arXiv preprint arXiv:1207.0580 (2012).
    1. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3:32–35. doi: 10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>;2-3.
    1. Altman NS. An introduction to kernel and nearest-neighbor nonparametric regression. The American Statistician. 1992;46:175–185.

Source: PubMed

Sponsors and Collaborators

Medical Conditions

Drug Interventions

3
Subscribe