Amaç: Bu çalışmada; yapay sinir ağlarının (YSA), panoramik filmlerde mandibular kondil morfolojisini belirlemedeki performansının değerlendirilmesi amaçlanmıştır. Gereç ve Yöntemler: Çalışma için 18 yaş altı bireylere ait olan toplam 1.645 dijital panoramik görüntü incelendi. Bu görüntüler üzerinde sağ ve sol eklem olmak üzere toplam 3.290 mandibular kondil bölgesi kesilerek morfolojik açıdan değerlendirildi. Kesilen görüntüler normal ve anormal olarak etiketlenen kondil görüntüleri YSA modeline verilmek üzere %75 eğitim seti, %15 doğrulama seti ve %10 test seti olarak ayrıldı. Çalışmada, sinir ağı mimarisi olarak DenseNet mimarisi kullanıldı. Bulgular: Çalışma kapsamında, özellikle seçilen sinir ağı modeli ile eğitim aşaması için %91,76'ya ulaşırken, test aşaması için %89,00 doğruluk oranı ile yüksek performansa ulaştığı varsayılmıştır. Buna göre normal sınıfı için 197 adet normal etiketi test edilirken, 19 adet veride yanlış olarak anormal etiketi bulunmuştur. Bununla birlikte değerlendirme sırasında 96 adet anormal sınıfı doğru olarak test edilirken 17 adet veri ise normal olarak değerlendirilmiştir. Sonuç: Mandibular kondil morfolojisi, YSA kullanılarak yüksek oranda doğru tespit edilmiştir. İlerde yapılacak çalışmalarda veri sayısı artırılarak başarının daha da artırılması mümkündür. Temporomandibular eklem bölgesinin yapay zekâ destekli programlar tarafından yüksek doğrulukla değerlendirilebilmesi, klinikte çokça karşılaşılan bu grup hastaların doğru tanı almalarını hızlandıracak ve doğru yönlendirme ile daha çabuk tedavi imkânı bulmalarını kolaylaştıracaktır.
Anahtar Kelimeler: Panoramik radyografi; mandibular kondil; yapay zekâ; bilgisayarlı görü teknikleri
Objective: In this study; it is aimed to evaluate the performance of artificial neural networks in determining the morphology of the mandibular condyle in panoramic images. Material and Methods: A total of 3,290 including right and left joints, mandibular condyles were cut and morphologically examined on 1,645 digital panoramic images. Condyle images that labeled as normal and abnormal were divided into 75% training set, 15% validation set and 10% test set to be given to the artificial neural network model. In this study, DenseNet architecture and GoogLeNet architectures was used as the neural network architecture. Results: Within the scope of the study, it was assumed that while it reached 91.76% for the training phase with the selected neural network model, it reached a high performance with an accuracy rate of 89.00% for the test phase. Accordingly, while 197 normal labels were tested for the normal class, 19 incorrectly abnormal labels were found in the data. However, during the evaluation, 96 classes of abnormal were tested correctly, while 17 data were evaluated as normal. Conclusion: A high rate of success has been achieved with the use of artificial neural networks in determining the morphology of the mandibular condyle. It is possible to further increase the success by increasing the number of data in future studies. Evaluating the temporomandibular joint region with high accuracy by artificial intelligence-supported programs will accelerate the correct diagnosis of this group of patients, who are frequently encountered in the clinic, and make it easier for them to get treatment more quickly with the right guidance.
Keywords: Panoramic radiography; mandibular condyle; artificial intelligence; computer vision techniques
- Mallya SM, Lam EWN. Radiographic anatomy. White and Pharoah's Oral Radiology. 8th ed. St. Louis, Missouri: Elsevier; 2019. p.504.
- Hegde S, Praveen B, Shetty SR. Morphological and radiological variations of mandibular condyles in health and diseases: a systematic review. Dentistry. 2013;3(1):154-8.
- Schwendicke F, Golla T, Dreher M, Krois J. Convolutional neural networks for dental image diagnostics: A scoping review. J Dent. 2019;91:103226. [Crossref] [PubMed]
- Géron A. Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow. 3rd ed. Sebastapol, CA: O'Reilly Media, Inc.; 2022.
- Ba-Hattab R, Barhom N, Osman SAA, Naceur I, Odeh A, Arisha Asad A, et al. Detection of periapical lesions on panoramic radiographs using deep learning. Applied Sciences. 2023;13(3):1516. [Crossref]
- Yaren Tekin B, Ozcan C, Pekince A, Yasa Y. An enhanced tooth segmentation and numbering according to FDI notation in bitewing radiographs. Comput Biol Med. 2022;146:105547. [Crossref] [PubMed]
- De Tobel J, Radesh P, Vandermeulen D, Thevissen PW. An automated technique to stage lower third molar development on panoramic radiographs for age estimation: a pilot study. J Forensic Odontostomatol. 2017;35(2):42-54. [PubMed] [PMC]
- Lee JH, Kim DH, Jeong SN. Diagnosis of cystic lesions using panoramic and cone beam computed tomographic images based on deep learning neural network. Oral Dis. 2020;26(1):152-8. [Crossref] [PubMed]
- Murata M, Ariji Y, Ohashi Y, Kawai T, Fukuda M, Funakoshi T, et al. Deep-learning classification using convolutional neural network for evaluation of maxillary sinusitis on panoramic radiography. Oral Radiol. 2019;35(3):301-7. [Crossref] [PubMed]
- De Brébisson A, Vincent P. An exploration of softmax alternatives belonging to the spherical loss family. arXiv. 2015;151105042. [Link]
- Vincent P, De Brébisson A, Bouthillier X. Efficient exact gradient update for training deep networks with very large sparse targets. arXiv. 2015;1412.7091. [Link]
- Yaohua X, Xudong M. A sar oil spill image recognition method based on densenet convolutional neural network. IEEE. 2019;78-81. [Crossref]
- Huang G, Liu Z, Van Der Maaten L, Weinberger KQ. Densely connected convolutional networks. IEEE. 2017;4700-8. [Crossref]
- Goutte C, Gaussier E. A probabilistic interpretation of precision, recall and F-score, with implication for evaluation. In: Losada DE, Fernández-Luna JM, eds. Advances in Information Retrieval. 1st ed. Berlin, Heidelberg: Springer; 2005. p.345-59. [Crossref]
- Kingma DP, Ba J. Adam: A Method for Stochastic Optimization. arXiv. 2014;1412.6980v9 2014. [Link]
- Honey OB, Scarfe WC, Hilgers MJ, Klueber K, Silveira AM, Haskell BS, et al. Accuracy of cone-beam computed tomography imaging of the temporomandibular joint: comparisons with panoramic radiology and linear tomography. Am J Orthod Dentofacial Orthop. 2007;132(4):429-38. [Crossref] [PubMed]
- Oliveira-Santos C, Bernardo RT, Capelozza ALÁ. Mandibular condyle morphology on panoramic radiographs of asymptomatic temporomandibular joints. IJD International Journal of Dentistry. 2009;8(3). [Link]
- Çamlıdağ İ, Sayıt AT, Elmalı M. Is condyle morphology a factor for anterior temporomandibular disc displacement? Turk J Med Sci. 2022;52(5):1609-15. [Crossref] [PubMed] [PMC]
- Yale SH, Ceballos M, Kresnoff CS, Hauptfuehrer JD. Some observations on the classification of mandibular condyle types. Oral Surg Oral Med Oral Pathol. 1963;16:572-7. [Crossref] [PubMed]
- Khanal P, Pranaya K. Study of mandibular condyle morphology using orthopantomogram. Journal of Nepal Dental Association. 2020;20(1):3-7. [Link]
- Hussain M, Bird JJ, Faria DR. A study on CNN transfer learning for image classification. In: Lotfi A, Bouchachia H, Gegov A, Langensiepen C, McGinnity M, eds. Advances in Computational Intelligence Systems. 1st ed. Switzerland AG: Springer International Publishing; 2019. p.191-202.
- Gupta J, Pathak S, Kumar G. Deep Learning (CNN) and Transfer Learning: A Review. Journal of Physics: Conference Series. 2022;2273(1):012029. [Crossref]
- Han D, Liu Q, Fan W. A new image classification method using CNN transfer learning and web data augmentation. Expert Systems with Applications. 2018;95:43-56. [Crossref]
- Kazangirler CB, Özcan C, Tekin BY. Ön Eğitimli Evrişimli Sinir Ağları ile UI Öğelerinin Tespiti ve Sınıflandırılması. 5th International Conference on Data Science and Applications (ICONDATA'22); 7-11 Ekim 2022; Muğla: Icondata; 2022. p.100-7. [Link]
- Kim D, Choi E, Jeong HG, Chang J, Youm S. Expert system for mandibular condyle detection and osteoarthritis classification in panoramic imaging using r-cnn and cnn. Applied Sciences. 2020;10(21):7463-4. [Crossref]
.: Process List