DERİN ÖĞRENMEYLE GELİŞTİRİLEN YAPAY ZEKA ALGORİTMALARIYLA LATERAL SEFALOMETRİK GÖRÜNTÜLER ÜZERİNDEN FARİNGEAL HAVA YOLUNUN DEĞERLENDİRİLMESİ

Abstract

ÖZET Amaç: Bu çalışmanın amacı, konik ışınlı bilgisayarlı tomografi görüntülerinden elde edilen lateral sefalometrik görüntüler üzerinde özel bir yapay zeka algoritması kullanılarak faringeal hava yolu tespitinin başarısını araştırmaktır. Gereç ve Yöntemler: Çalışmamızın veri seti, özel bir yapay zeka algoritması kullanılarak 1040 hastanın ortodontik tedavi öncesi konik ışınlı bilgisayarlı tomografi görüntülerinden elde edilen lateral sefalometrik radyografiler üzerinde gerçekleştirildi ve serbest çizim tekniği ile segmentasyon yöntemi uygulandı ve faringeal hava yolu belirlendi. Görüntüler üzerindeki hava yolu etiketlemesi CranioCatch yapay zeka yazılımı (CranioCatch, Eskisehir, Türkiye) kullanılarak yapıldı. Bulgular: Yapay zeka modeli Yolov5x modeli ile 500 epoch ve 0,01 öğrenme oranıyla eğitildi. Çalışmada eğitilen yapay zeka modelinde duyarlılık, kesinlik ve F1 puanları sırasıyla 1, 0,9903 ve 0,9951 olarak gerçekleşti. Sonuç: Faringeal hava yolunu değerlendirdiğimiz model genel olarak başarılıydı. Çalışmamız gelecekteki KIBT raporlama sistemlerinin geliştirilmesi açısından umut vericidir. Derin öğrenmeye dayalı bu sistemlerin rutin klinik uygulamalarda karar destek mekanizması olarak hekimlere zaman kazandıracağı düşünülmektedir. Ayrıca faringeal hava yolunun değerlendirilmesinde gözlemciler arası farklılıkların ve gözlemcilerin farklı zamanlarda yaptığı değerlendirmelerde oluşabilecek tutarsızlıkların en aza indirilmesine yardımcı olacağı öngörülmektedir.

Keywords

• Yapay zeka
• CBCT
• Derin öğrenme
• Faringeal havayolu

EVALUATION OF THE PHARYNGEAL AIRWAY WITH ARTIFICIAL INTELLIGENCE ALGORITHMS DEVELOPED BY DEEP LEARNING FROM LATERAL CEPHALOMETRIC IMAGE

Abstract

ABSTRACT Objectives: The aim of this study is to investigate the success of pharyngeal airway detection using a special artificial intelligence algorithm on lateral cephalometric images obtained from cone beam computed tomography images. Materials and Methods: The data set of our study was performed on the lateral cephalometric radiographs was obtained from cone beam computed tomography images of 1040 patients before orthodontic treatment using a special artificial intelligence algorithm and the segmentation method were applied with the free drawing tchnique and the pharyngeal airway was determined. Airway labeling on images was done using CranioCatch annotation software (CranioCatch, Eskişehir, Turkey). Results: The artificial intelligence model was trained with the Yolov5x model as 500 epochs and 0.01 learning rate. Sensitivity, precision and F1 scores in the artifical intelligence model trained in the study were 1, 0.9903 and 0.9951 respectively. Conclusion: The model in which we evaluated the pharyngeal airway was generally successful. Our study is promising for the development of future CBCT reporting systems. It is thought that these deep learning-based systems will save physicians time as a decision support mechanism in routine clinical practices. It is also anticipated that it will help in minimizing interobserver differences in the evaluation of the pharyngeal airway and inconsistencies that may occur in the evaluations made by observers at different times.

Keywords

• Artificial Intelligence
• CBCT
• Deep learning
• Pharyngeal airway

References

1. Sahoo NK, Jayan B, Ramakrishna N, Chopra SS, Kochar G. Evaluation of upper airway dimensional changes and hyoid position following mandibular advancement in patients with skeletal class II malocclusion. J Craniofac Surg. 2012;23(6):e623-e7.
2. Angle EH. Treatment of malocclusion of the teeth: Angle's system: SS White Dental Mfg Co; 1907.
3. Guilleminault C. Obstructive sleep apnea: the clinical syndrome and historical perspective. Med. Clin. N. Am. 1985;69(6):1187-203.
4. Allen Jr B, Seltzer SE, Langlotz CP, Dreyer KP, Summers RM, Petrick N, et al. A road map for translational research on artificial intelligence in medical imaging: from the 2018 National Institutes of Health/RSNA/ACR/The Academy Workshop. J. Am. Coll. Radiol. 2019;16(9):1179-89.
5. Sen D, Chakrabarti R, Chatterjee S, Grewal D, Manrai K. Artificial intelligence and the radiologist: the future in the Armed Forces Medical Services. BMJ Mil Health. 2020;166(4):254-6.
6. Yu H, Cho S, Kim M, Kim W, Kim J, Choi J. Automated skeletal classification with lateral cephalometry based on artificial intelligence. J. Dent. Res. 2020;99(3):249-56.
7. Aboudara C, Hatcher D, Nielsen I, Miller A. A threedimensional evaluation of the upper airway in adolescents. Orthod & Craniofac Res. 2003;6:173-5.
8. Kök H, Acilar AM, İzgi MS. Usage and comparison of artificial intelligence algorithms for determination of growth and development by cervical vertebrae stages in orthodontics. Prog Orthod. 2019;20:1-10.

9. Pinchi V, Pradella F, Vitale G, Rugo D, Nieri M, Norelli G-A. Comparison of the diagnostic accuracy, sensitivity and specificity of four odontological methods for age evaluation in Italian children at the age threshold of 14 years using ROC curves. Med Sci Law. 2016;56(1):13-8.
10. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin. Chem. 1993;39(4):561-77.
11. Davis J, Goadrich M, editors. The relationship between Precision-Recall and ROC curves. Proceedings of the 23rd International Conference on Machine Learning; 2006.
12. Arat M, Iseri H, Iseri V. İskeletsel açık kapanışa yol açan faktörlerin sagittal yüz yapısına gore incelenmesi. Turk J Orthod. 1996;9:155-62.
13. Becker OE, Avelar RL, Göelzer JG, do Nascimento Dolzan A, Júnior OLH, De Oliveira RB. Pharyngeal airway changes in class III patients treated with double jaw orthognathic surgery-maxillary advancement and mandibular setback. J. Maxillofac. Surg. 2012;70(11):e639-e47.
14. Choi S-K, Yoon J-E, Cho J-W, Kim J-W, Kim S-J, Kim M-R. Changes of the airway space and the position of hyoid bone after mandibular set back surgery using bilateral sagittal split ramus osteotomy technique. Maxillofac Plast Reconstr Surg. 2014;36(5):185.
15. Jakobsone G, Stenvik A, Espeland L. The effect of maxillary advancement and impaction on the upper airway after bimaxillary surgery to correct Class III malocclusion. Am J Orthod Dentofacial Orthop. 2011;139(4):e369-e76.
16. Preston CB, Lampasso JD, Tobias PV, editors. Cephalometric evaluation and measurement of the upper airway. Semin. Orthod; 2004: Elsevier.
17. Moon J-H, Hwang H-W, Yu Y, Kim M-G, Donatelli RE, Lee S-J. How much deep learning is enough for automatic identification to be reliable? A cephalometric example. The Angle Orthod. 2020;90(6):823-30.
18. Sin Ç, Akkaya N, Aksoy S, Orhan K, Öz U. A deep learning algorithm proposal to automatic pharyngeal airway detection and segmentation on CBCT images. Orthod Craniofac Res. 2021;24:117-23.
19. Kim M-J, Liu Y, Oh SH, Ahn H-W, Kim S-H, Nelson G. Automatic cephalometric landmark identification system based on the multi-stage convolutional neural networks with CBCT combination images. Sensors. 2021;21(2):505.
20. Leonardi R, Giudice AL, Farronato M, Ronsivalle V, Allegrini S, Musumeci G, et al. Fully automatic segmentation of sinonasal cavity and pharyngeal airway based on convolutional neural networks. Am J Orthod Dentofacial Orthop. 2021;159(6):824-35.