Investigation of The Effect on Model Parameters in SegFormer Method for Tooth Segmentation

Yıl 2025, Cilt: 8 Sayı: 1, 132 - 141, 15.01.2025

Egemen Belge , Seçkin Karasu

https://doi.org/10.34248/bsengineering.1569645

Öz

In recent years, AI-based applications in computer-aided treatments have been increasingly used by dentists for disease diagnosis. Accurate segmentation is crucial in the process of identifying dental diseases. Manual segmentation of teeth significantly increases the time and workload for dentists when making diagnoses. At this stage, the automatic segmentation of the dental region using machine learning and artificial intelligence methods has become a topic of great interest for researchers. In the study, X-Ray tooth ratios taken from 598 patients with 15318 polygons by 12 dental interns are used. The dataset is split into 70% for training, 15% for validation, and 15% for testing. This dataset is used in the training process of a deep learning network for automatic tooth segmentation. The performances of architectures generated based on changes in the hyperparameters of the SegFormer training block are examined. Here, according to the Dice similarity coefficients of the models according to the Mix BO-B5 architectures, the performances for the test data are obtained as 92.61%, 92.82%, 93.25%, 93.13%, 93.17% and 93.09%, respectively. According to the test results obtained, the developed artificial intelligence-based SegFormer network performs tooth segmentation with high accuracy. The developed deep learning network can be efficiently used especially in the diagnosis of dental diseases. High Dice similarity coefficients indicate that the SegFormer network presented in this study can accurately detect the tooth region.

Anahtar Kelimeler

Artificial intelligence, Tooth segmentation, SegFormer, X-Ray imaging, Dice similarity coefficient

Etik Beyan

Bu araştırmada hayvanlar ve insanlar üzerinde herhangi bir çalışma yapılmadığı için etik kurul onayı alınmamıştır.

Diş Segmentasyonunda Segformer Yönteminin Model Parametreleri Üzerindeki Etkisinin Araştırılması

Öz

Son yıllarda bilgisayar destekli tedavilerde yapay zekâ temelli uygulamalar diş hekimleri için hastalık teşhisinde kullanımı giderek artmaktadır. Diş hastalığın belirlenmesi sürecinde doğru bir şekilde bölütleme işlemi oldukça önemlidir. Diş bölütlenmesinin manuel olarak yapılması diş hekimleri tarafından yapılan teşhis için geçen süreyi ve işlem yükünü önemli ölçüde arttırmaktadır. Bu aşamada, makine öğrenmesi ve yapay zekâ yöntemleriyle otomatik olarak diş bölgesinin bölütlenmesi araştırmacıların ilgi duyduğu güncel bir konu olmaktadır. Çalışmada 12 diş stajyeri tarafından 15318 poligonlu 598 hastadan alınan X-Ray diş görüntüleri kullanılmaktadır. Kullanılan veri seti eğitim, doğrulama ve test olarak %70, %15, %15 olarak bölünmüştür. Bu veri seti otomatik olarak diş bölütlemeyi amaçlayan derin öğrenme ağının eğitim sürecinde kullanılmaktadır. SegFormer eğitim bloğu hiper parametrelerinin değişimine bağlı oluşturulan mimarilerinin performansları incelenmektedir. Burada MiT BO-B5 mimarilerine göre oluşturulan modellerin Dice benzerlik katsayılarına göre test verisi için performansları sırasıyla %92,61, %92,82, %93,25, %93,13, %93,17 ve %93,09 olarak elde edilmektedir. Elde edilen test sonuçlarına göre geliştirilen yapay zekâ tabanlı SegFormer ağı diş bölütlemeyi yüksek doğrulukla gerçekleştirmektedir. Geliştirilen derin öğrenme ağı özellikle diş hastalıklarının teşhisinde girdi olarak verimli bir şekilde kullanılabilecektir. Yüksek Dice benzerlik katsayıları, çalışmada sunulan SegFormer ağının diş bölgesini doğru bir şekilde tespit edebildiğini ifade etmektedir.

Anahtar Kelimeler

Yapay zekâ, Diş bölütlenmesi, SegFormer, X-Ray görüntüleme, Dice benzerlik katsayısı

Etik Beyan

Bu araştırmada hayvanlar ve insanlar üzerinde herhangi bir çalışma yapılmadığı için etik kurul onayı alınmamıştır.

Kaynakça

• Alam MK, Haque T, Akhte, F, Albagieh HN, Nabhan AB, Alsenani MA, Islam S. 2023. Teeth segmentation by optical radiographic images using VGG-16 deep learning convolution architecture with R-CNN network approach for biomedical sensing applications. Optic Quant Electron, 55(9): 808.
• Chen H, Zhang K, Lyu P, Li H, Zhang L, Wu J, Lee CH. 2019. A deep learning approach to automatic teeth detection and numbering based on object detection in dental periapical films. Sci Rep, 9(1): 3840.
• Gao C, Meng D, Yang Y, Wang Y, Zhou X, Hauptmann AG. 2013. Infrared patch-image model for small target detection in a single image. IEEE Transact Image Proces, 22(12): 4996-5009.
• Gong Y, Zhang J, Cheng J, Yuan W, He L. 2024. Automatic tooth segmentation for patients with alveolar clefts guided by tooth descriptors. Biomed Signal Proces Cont, 90: 105821.
• He K, Gkioxari G, Dollár P, Girshick R. 2017. Mask R-CNN. In Proc IEEE Int Conf Comput Vision, 2017: 2961-2969.
• Humans In The Loop. 2023. Teeth Segmentation on dental X-ray images [Data set]. Kaggle.
• Indraswari R, Arifin AZ, Navastara DA, Jawas N. 2015. Teeth segmentation on dental panoramic radiographs using decimation-free directional filter bank thresholding and multistage adaptive thresholding. International Conference on Information & Communication Technology and Systems (ICTS), September 16, Surabaya, Indonesia, pp: 49-54.
• Kang HC, Choi C, Shin J, Lee J, Shin YG. 2015. Fast and accurate semiautomatic segmentation of individual teeth from dental CT images. Computat Math Meth Medic, 2015(1): 810796.
• Kato S, Hotta K. 2024. Adaptive t-vMF dice loss: An effective expansion of dice loss for medical image segmentation. Comput Biol Medic, 168: 107695.
• Lee SJ, Chung D, Asano A, Sasaki D, Maeno M, Ishida Y, Kobayashi T, Kuwajima Y, Silva JDD, Nagai S. 2022. Diagnosis of tooth prognosis using artificial intelligence. Diagnostics, 12(6): 1422.
• Li H, Sun G, Sun H, Liu W. 2012. Watershed algorithm based on morphology for dental x-ray images segmentation. International Conference on Signal Processing, August 12-15, Hong Kong, China, pp: 877-880.
• Li Z, Wang H. 2016. Interactive tooth separation from dental model using segmentation field. PloS One, 11(8): e0161159.
• Lira PHM, Giraldi GA, Neves LA. 2009. An automatic morphometrics data extraction method in dental x-ray image. International Conference on Biodental Enginnering, June 26-27, Porto, Portugal, pp: 77-82.
• Meng D, Zhao Q, Jiang L. 2017. A theoretical understanding of self-paced learning. Info Sci, 414: 319-328.
• Modi CK, Desai NP. 2011. A simple and novel algorithm for automatic selection of ROI for dental radiograph segmentation. Canadian Conference on Electrical and Computer Engineering, May 08-11, Niagara Falls, Canada, pp: 000504-000507. https://doi.org/10.1109/CCECE.2011.6030501.
• Nagaraju P, Sudha SV. 2024. Design of a novel panoptic segmentation using multi-scale pooling model for tooth segmentation. Soft Comput, 28(5): 4185-4196.
• Poonsri A, Aimjirakul N, Charoenpong T, Sukjamsri C. 2016. Teeth segmentation from dental x-ray image by template matching. 9th Biomedical Engineering International Conference, December 7-9, Laung Prabang, Laos, pp: 1-4.
• Prajapati SA, Nagaraj R, Mitra S. 2017. Classification of dental diseases using CNN and transfer learning. 5th International Symposium on Computational and Business Intelligence, August 11-14, Dubai, United Arab Emirates, pp: 70-74.
• Said EH, Nassar DEM, Fahmy G, Ammar HH. 2006. Teeth segmentation in digitized dental X-ray films using mathematical morphology. IEEE Transact Info Forens Secur, 1(2): 178-189.
• Sheng C, Wang L, Huang Z, Wang T, Guo Y, Hou W, Yan X. 2023. Transformer-based deep learning network for tooth segmentation on panoramic radiographs. J Syst Sci Compl, 36(1): 257-272.
• Tian S, Dai N, Zhang B, Yuan F, Yu Q, Cheng X. 2019. Automatic classification and segmentation of teeth on 3D dental model using hierarchical deep learning networks. IEEE Access, 7: 84817-84828.
• Wang CW, Huang CT, Lee JH, Li CH, Chang SW, Siao MJ, Lai TM, Ibragimov B, Vrtovec T, Ronneberger O, Fischer P, Cootes TF, Lindner C. 2016. A benchmark for comparison of dental radiography analysis algorithms. Medic Image Analy, 31: 63-76.
• Wirtz A, Mirashi SG, Wesarg S. 2018. Automatic teeth segmentation in panoramic X-ray images using a coupled shape model in combination with a neural network. In Medical Image Computing and Computer Assisted Intervention–MICCAI 2018: 21st International Conference, September 16-20, Granada, Spain, Part IV(11): 712-719.
• Wu A, Zhu L, Han Y, Yang Y. 2019. Connective cognition network for directional visual commonsense reasoning. Adv Neural Info Proces Syst, 32.
• Xie E, Wang W, Yu Z, Anandkumar A, Alvarez JM, Luo P. 2021. SegFormer: Simple and efficient design for semantic segmentation with transformers. Adv Neural Info Proces Syst, 34: 12077-12090.
• Xu M, Wu Y, Xu Z, Ding P, Bai H, Deng, X. 2023. Robust automated teeth identification from dental radiographs using deep learning. J Dentistry, 136: 104607.
• Yan M, Guo J, Tian W, Yi Z. 2018. Symmetric convolutional neural network for mandible segmentation. Know Based Syst, 159: 63-71.
• Yang J, Xie Y, Liu L, Xia B, Cao Z, Guo C. 2018. Automated dental image analysis by deep learning on small dataset. 42nd Annual Computer Software and Applications Conference (COMPSAC), July 23-27, Tokyo, Japan, pp: 492-497.
• Yong H, Meng D, Zuo W, Zhang L. 2017. Robust online matrix factorization for dynamic background subtraction. IEEE Transact Pattern Analy Machine Intel, 40(7): 1726-1740.
• Zhao Y, Li P, Gao C, Liu Y, Chen Q, Yang F, Meng D. 2020. TSASNet: Tooth segmentation on dental panoramic X-ray images by Two-Stage Attention Segmentation Network. Know Based Syst, 206: 106338
• Zhu L, Yang Y. 2020. Actbert: Learning global-local video-text representations. IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 13-19, Seattle, US, pp: 8746-8755.