Imaging Science in Dentistry

Korean Academy of Oral and Maxillofacial Radiology (대한영상치의학회)
권호 표지
DOI QR코드

DOI QR Code

Clinical validity and precision of deep learning-based cone-beam computed tomography automatic landmarking algorithm

  • Jungeun Park (Department of Orthodontics, College of Dentistry, Yonsei University) ;
  • Seongwon Yoon (College of Dentistry, Seoul National University) ;
  • Hannah Kim (Imagoworks Incorporated) ;
  • Youngjun Kim (Imagoworks Incorporated) ;
  • Uilyong Lee (Department of Oral and Maxillofacial Surgery, College of Dentistry, Chungang University Hospital) ;
  • Hyungseog Yu (Department of Orthodontics, The Institute of Craniofacial Deformity, College of Dentistry, Yonsei University)
  • Received : 2024.01.22
  • Accepted : 2024.05.24
  • Published : 2024.09.30
Download PDF
⟨ Previous Next ⟩

Abstract

Purpose: This study was performed to assess the clinical validity and accuracy of a deep learning-based automatic landmarking algorithm for cone-beam computed tomography (CBCT). Three-dimensional (3D) CBCT head measurements obtained through manual and automatic landmarking were compared. Materials and Methods: A total of 80 CBCT scans were divided into 3 groups: non-surgical (39 cases); surgical without hardware, namely surgical plates and mini-screws (9 cases); and surgical with hardware (32 cases). Each CBCT scan was analyzed to obtain 53 measurements, comprising 27 lengths, 21 angles, and 5 ratios, which were determined based on 65 landmarks identified using either a manual or a 3D automatic landmark detection method. Results: In comparing measurement values derived from manual and artificial intelligence landmarking, 6 items displayed significant differences: R U6CP-L U6CP, R L3CP-L L3CP, S-N, Or_R-R U3CP, L1L to Me-GoL, and GoR-Gn/S-N (P<0.05). Of the 3 groups, the surgical scans without hardware exhibited the lowest error, reflecting the smallest difference in measurements between human- and artificial intelligence-based landmarking. The time required to identify 65 landmarks was approximately 40-60 minutes per CBCT volume when done manually, compared to 10.9 seconds for the artificial intelligence method (PC specifications: GeForce 2080Ti, 64GB RAM, and an Intel i7 CPU at 3.6 GHz). Conclusion: Measurements obtained with a deep learning-based CBCT automatic landmarking algorithm were similar in accuracy to values derived from manually determined points. By decreasing the time required to calculate these measurements, the efficiency of diagnosis and treatment may be improved.

Keywords

Acknowledgement

This work was supported by the Korea Medical Device Development Fund grant, funded by the Korean government (Ministry of Science and ICT; Ministry of Trade, Industry and Energy; Ministry of Health & Welfare; Ministry of Food and Drug Safety) under Project No. RS-2020-KD000002.

References

  1. Gribel BF, Gribel MN, Frazao DC, McNamara JA Jr, Manzi FR. Accuracy and reliability of craniometric measurements on lateral cephalometry and 3D measurements on CBCT scans. Angle Orthod 2011; 81: 26-35. https://doi.org/10.2319/032210-166.1
  2. Lee SH, Kil TJ, Park KR, Kim BC, Kim JG, Piao Z, et al. Three-dimensional architectural and structural analysis - a transition in concept and design from Delaire's cephalometric analysis. Int J Oral Maxillofac Surg 2014; 43: 1154-60. https://doi.org/10.1016/j.ijom.2014.03.012
  3. Olszewski R, Cosnard G, Macq B, Mahy P, Reychler H. 3D CT-based cephalometric analysis: 3D cephalometric theoretical concept and software. Neuroradiology 2006; 48: 853-62. https://doi.org/10.1007/s00234-006-0140-x
  4. Mah JK, Huang JC, Choo H. Practical applications of conebeam computed tomography in orthodontics. J Am Dent Assoc 2010; 141 Suppl 3: 7S-13.
  5. Lindner C, Wang CW, Huang CT, Li CH, Chang SW, Cootes TF. Fully automatic system for accurate localisation and analysis of cephalometric landmarks in lateral cephalograms. Sci Rep 2016; 6: 33581.
  6. Vandaele R, Aceto J, Muller M, Peronnet F, Debat V, Wang CW, et al. Landmark detection in 2D bioimages for geometric morphometrics: a multi-resolution tree-based approach. Sci Rep 2018; 8: 538.
  7. Lagravere MO, Low C, Flores-Mir C, Chung R, Carey JP, Heo G, et al. Intraexaminer and interexaminer reliabilities of landmark identification on digitized lateral cephalograms and formatted 3-dimensional cone-beam computerized tomography images. Am J Orthod Dentofacial Orthop 2010; 137: 598-604. https://doi.org/10.1016/j.ajodo.2008.07.018
  8. Hassan B, Nijkamp P, Verheij H, Tairie J, Vink C, van der Stelt P, et al. Precision of identifying cephalometric landmarks with cone beam computed tomography in vivo. Eur J Orthod 2013; 35: 38-44. https://doi.org/10.1093/ejo/cjr050
  9. Ghowsi A, Hatcher D, Suh H, Wile D, Castro W, Krueger J, et al. Automated landmark identification on cone-beam computed tomography: accuracy and reliability. Angle Orthod 2022; 92: 642-54. https://doi.org/10.2319/122121-928.1
  10. Jeon S, Lee KC. Comparison of cephalometric measurements between conventional and automatic cephalometric analysis using convolutional neural network. Prog Orthod 2021; 22: 14.
  11. Gupta A, Kharbanda OP, Sardana V, Balachandran R, Sardana HK. Accuracy of 3D cephalometric measurements based on an automatic knowledge-based landmark detection algorithm. Int J Comput Assist Radiol Surg 2016; 11: 1297-309. https://doi.org/10.1007/s11548-015-1334-7
  12. Barrett JF, Keat N. Artifacts in CT: recognition and avoidance. Radiographics 2004; 24: 1679-91. https://doi.org/10.1148/rg.246045065
  13. Hung K, Yeung AW, Tanaka R, Bornstein MM. Current applications, opportunities, and limitations of AI for 3D imaging in dental research and practice. Int J Environ Res Public Health 2020; 17: 4424.
  14. Torosdagli N, Liberton DK, Verma P, Sincan M, Lee JS, Bagci U. Deep geodesic learning for segmentation and anatomical landmarking. IEEE Trans Med Imaging 2019; 38: 919-31. https://doi.org/10.1109/TMI.2018.2875814
  15. Lee JH, Yu HJ, Kim MJ, Kim JW, Choi J. Automated cephalometric landmark detection with confidence regions using Bayesian convolutional neural networks. BMC Oral Health 2020; 20: 270.
  16. Makram M, Kamel H. Reeb graph for automatic 3D cephalometry. Int J Image Process 2014; 8: 17-29.
  17. Montufar J, Romero M, Scougall-Vilchis RJ. Hybrid approach for automatic cephalometric landmark annotation on conebeam computed tomography volumes. Am J Orthod Dentofacial Orthop 2018; 154: 140-50. https://doi.org/10.1016/j.ajodo.2017.08.028
  18. Minnema J, van Eijnatten M, Hendriksen AA, Liberton N, Pelt DM, Batenburg KJ, et al. Segmentation of dental conebeam CT scans affected by metal artifacts using a mixed-scale dense convolutional neural network. Med Phys 2019; 46: 5027-35. https://doi.org/10.1002/mp.13793
  19. Dot G, Rafflenbeul F, Arbotto M, Gajny L, Rouch P, Schouman T. Accuracy and reliability of automatic three-dimensional cephalometric landmarking. Int J Oral Maxillofac Surg 2020; 49: 1367-78. https://doi.org/10.1016/j.ijom.2020.02.015