All-on-4®全口速定植牙治療方案已盛行於全口無牙患者的療程中，療程時間短、成功率高以及避免補骨手術的風險。然而，臨床醫師在手術過程中通常依賴經驗判斷患者的顎骨狀況，缺乏統一的標準或決策建議。本研究的設計流程旨在提供一個結合機器學習和生物力學分析以及最佳化方法於臨床醫療決策的完整系統，在進行All-on-4®全口速定植牙之臨床治療時能提供具即時性的設計參考。
本研究分成兩個部分，第一部分為利用粒子點群法結合人工神經網路 (Artificial Neural Network, ANN)模型，搜尋植體最佳配置以降低植體周圍骨骼的平均應力值。第二部分為訓練基於雙向演進式結構最佳化 (Bi-directional Evolutionary Structural Optimization, BESO)的卷積神經網路 (Convolutional Neural Networks, CNN)進行支架最佳化省去BESO耗時的疊代演化過程，在保持一定的結構剛性下節省硬材料的使用。本研究提出之方法結合機器學習、有限元素分析以及最佳化方法，可應用於不同外形之下顎骨，快速找尋最適合的All-on-4®設計。
以ANN模型結合最佳化方法搜尋結果顯示，六種顎骨模型(STD、DH、RH、MH、LH、ML) 最佳植體配置之的平均應力目標函數分別下降10.42 %、10.39 %、2.71 %、20.8 %、18.14 %、27.3 %，與有限元素分析結果比較，六種模型的平均誤差為6.08 %。以CNN模型進行無迭代的BESO結果與傳統BESO結合有限元素分析相比，六種顎骨模型所得之支架最佳設計之平均順從性誤差為0.26 %，平均形狀誤差為10.02 %。
本研究結合機器學習與最佳化方法之設計流程，可針對不同幾何的下顎骨模型在5 ～ 10分鐘內完成基於生物力學的贋復物設計最佳化，能於臨床上給予醫師即時性植體配置建議。

The "All-on-4®" treatment approach, which is widely used for edentulous patients, involves supporting a fixed prosthesis with four dental implants. However, during the treatment planning process, clinicians often rely on their experience to assess the condition of the patient's jawbone and determine the implant positions and denture framework design. This process lacks biomechanical considerations. In this study, we propose a Convolutional Neural Network (CNN) model based on Bidirectional Evolutionary Structural Optimization (BESO), called BESO-Net, to directly predict the optimal prosthetic framework design without iterative evolution. In addition, BESO-Net ensures structural stiffness while minimizing the use of hard materials.
The BESO-Net approach combines machine learning, finite element analysis and optimization methods to efficiently customize the optimal prosthetic framework for complex 3D structures using multiple materials. It enables rapid customization of ideal prosthetic frameworks according to different mandibular shapes. To evaluate the performance of BESO-Net, we compared its results with those obtained using traditional BESO combined with finite element analysis. The average error in predicting the five optimal framework designs was found to be 0.29% for positional accuracy and 11.26% for shape accuracy. In particular, the computational time for structural optimization was significantly reduced from 6 hours 35 min to 5-10 minutes.
This study presents a novel design process that integrates machine learning and optimization techniques to provide biomechanically optimized prosthetic framework designs for different mandibular geometries. Clinicians can benefit from real-time prosthesis design recommendations based on biomechanical considerations, offering improved clinical outcomes in the All-on-4® treatment of edentulous patients.

[1] T. Liu, Z. X. Mu, T. Yu, C. Wang, and Y. D. Huang, "Biomechanical comparison of implant inclinations and load times with the all-on-4 treatment concept: a three-dimensional finite element analysis," Computer Methods in Biomechanics and Biomedical Engineering, vol. 22, no. 6, pp. 585-594, Apr 2019, doi: 10.1080/10255842.2019.1572120.
[2] 林家葳, "基於有限元素法的機器學習技術實現All-on-4®贋復設計之即時最佳化," 國立成功大學機械工程學系學位論文, 2022.
[3] O. Ozan and S. Kurtulmus-Yilmaz, "Biomechanical Comparison of Different Implant Inclinations and Cantilever Lengths in All-on-4 Treatment Concept by Three-Dimensional Finite Element Analysis," International Journal of Oral & Maxillofacial Implants, vol. 33, no. 1, pp. 64-71, Jan-Feb 2018, doi: 10.11607/jomi.6201.
[4] P. Malo, B. Rangert, and M. Nobre, "All-on-4 immediate-function concept with Branemark System (R) implants for completely edentulous maxillae: A 1-year retrospective clinical study," Clinical Implant Dentistry and Related Research, vol. 7, pp. S88-S94, 2005, doi: 10.1111/j.1708-8208.2005.tb00080.x.
[5] P. Malo, M. D. Nobre, A. Lopes, C. Francischone, and M. Rigolizzo, ""All-on-4" Immediate-Function Concept for Completely Edentulous Maxillae: A Clinical Report on the Medium (3 Years) and Long-Term (5 Years) Outcomes," Clinical Implant Dentistry and Related Research, vol. 14, pp. e139-e150, May 2012, doi: 10.1111/j.1708-8208.2011.00395.x.
[6] P. Malo, M. D. Nobre, A. Lopez, S. M. Moss, and G. J. Molina, "A longitudinal study of the survival of All-on-4 implants in the mandible with up to 10 Years of follow-up," Journal of the American Dental Association, vol. 142, no. 3, pp. 310-320, Mar 2011, doi: 10.14219/jada.archive.2011.0170.
[7] A. Lopes, P. Malo, M. D. Nobre, and E. Sanchez-Fernandez, "The NobelGuide (R) All-on-4 (R) Treatment Concept for Rehabilitation of Edentulous Jaws: A Prospective Report on Medium- and Long-Term Outcomes," Clinical Implant Dentistry and Related Research, vol. 17, pp. E406-E416, Oct 2015, doi: 10.1111/cid.12260.
[8] D. O. Dogan, N. T. Polat, S. Polat, E. Seker, and E. B. Gul, "Evaluation of "All-on-Four" Concept and Alternative Designs with 3D Finite Element Analysis Method," Clinical Implant Dentistry and Related Research, vol. 16, no. 4, pp. 501-510, Aug 2014, doi: 10.1111/cid.12024.
[9] C. Drago and K. Howell, "Concepts for Designing and Fabricating Metal Implant Frameworks for Hybrid Implant Prostheses," Journal of Prosthodontics-Implant Esthetic and Reconstructive Dentistry, vol. 21, no. 5, pp. 413-424, Jul 2012, doi: 10.1111/j.1532-849X.2012.00835.x.
[10] F. Haroun and O. Ozan, "Evaluation of Stresses on Implant, Bone, and Restorative Materials Caused by Different Opposing Arch Materials in Hybrid Prosthetic Restorations Using the All-on-4 Technique," Materials, vol. 14, no. 15, Aug 2021, Art no. 4308, doi: 10.3390/ma14154308.
[11] R. T. Abreu, A. O. Spazzin, P. Y. Noritomi, R. L. X. Consani, and M. F. Mesquita, "Influence of Material of Overdenture-Retaining Bar with Vertical Misfit on Three-Dimensional Stress Distribution," Journal of Prosthodontics-Implant Esthetic and Reconstructive Dentistry, vol. 19, no. 6, pp. 425-431, Aug 2010, doi: 10.1111/j.1532-849X.2010.00612.x.
[12] A. Bacchi, R. L. X. Consani, M. F. Mesquita, and M. B. F. dos Santos, "Stress distribution in fixed-partial prosthesis and peri-implant bone tissue with different framework materials and vertical misfit levels: a three-dimensional finite element analysis," Journal of Oral Science, vol. 55, no. 3, pp. 239-244, Sep 2013, doi: 10.2334/josnusd.55.239.
[13] S. Horita, T. Sugiura, K. Yamamoto, K. Murakami, Y. Imai, and T. Kirita, "Biomechanical analysis of immediately loaded implants according to the "All-on-Four" concept," Journal of Prosthodontic Research, vol. 61, no. 2, pp. 123-132, Apr 2017, doi: 10.1016/j.jpor.2016.08.002.
[14] C. H. Li, C. H. Wu, and C. L. Lin, "Design of a patient-specific mandible reconstruction implant with dental prosthesis for metal 3D printing using integrated weighted topology optimization and finite element analysis," Journal of the Mechanical Behavior of Biomedical Materials, vol. 105, May 2020, Art no. 103700, doi: 10.1016/j.jmbbm.2020.103700.
[15] J. Park, D. Lee, and A. Sutradhar, "Topology optimization of fixed complete denture framework," INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN BIOMEDICAL ENGINEERING, vol. 35, no. 6, JUN 2019, Art no. e3193, doi: 10.1002/cnm.3193.
[16] Z. P. Zhang, J. N. Chen, E. Li, W. Li, M. Swain, and Q. Li, "Topological design of all-ceramic dental bridges for enhancing fracture resistance," International Journal for Numerical Methods in Biomedical Engineering, vol. 32, no. 6, Jun 2016, Art no. e02749, doi: 10.1002/cnm.2749.
[17] K. J. Cheng et al., "Topological optimization of 3D printed bone analog with PEKK for surgical mandibular reconstruction," Journal of the Mechanical Behavior of Biomedical Materials, vol. 107, Jul 2020, Art no. 103758, doi: 10.1016/j.jmbbm.2020.103758.
[18] F. Mayta-Tovalino, A. Munive-Degregori, S. Luza, F. Cardenas-Marino, M. Guerrero, and J. Barja-Ore, "Applications and perspectives of artificial intelligence, machine learning and "dentronics" in dentistry: A literature review," Journal of International Society of Preventive and Community Dentistry, vol. 13, no. 1, pp. 1-8, Jan-Feb 2023, doi: 10.4103/jispcd.JISPCD_35_22.
[19] R. Phellan, B. Hachem, J. Clin, J. M. Mac-Thiong, and L. Duong, "Real-time biomechanics using the finite element method and machine learning: Review and perspective," Medical Physics, vol. 48, no. 1, pp. 7-18, Jan 2021, doi: 10.1002/mp.14602.
[20] N. Black and A. R. Najafi, "Learning finite element convergence with the Multi-fidelity Graph Neural Network?," Computer Methods in Applied Mechanics and Engineering, vol. 397, Jul 2022, Art no. 115120, doi: 10.1016/j.cma.2022.115120.
[21] H. Sasaki and H. Igarashi, "Topology Optimization Accelerated by Deep Learning," Ieee Transactions on Magnetics, vol. 55, no. 6, Jun 2019, Art no. 7401305, doi: 10.1109/tmag.2019.2901906.
[22] J. X. Luo, Y. Li, W. E. Zhou, Z. Q. Gong, Z. Y. Zhang, and W. Yao, "An Improved Data-Driven Topology Optimization Method Using Feature Pyramid Networks with Physical Constraints," Cmes-Computer Modeling in Engineering & Sciences, vol. 128, no. 3, pp. 823-848, 2021, doi: 10.32604/cmes.2021.016737.
[23] Y. Xing and L. Y. Tong, "A machine learning-assisted structural optimization scheme for fast-tracking topology optimization," Structural and Multidisciplinary Optimization, vol. 65, no. 4, Apr 2022, Art no. 105, doi: 10.1007/s00158-022-03181-5.
[24] S. Lee, H. Kim, Q. X. Lieu, and J. Lee, "CNN-based image recognition for topology optimization," Knowledge-Based Systems, vol. 198, Jun 2020, Art no. 105887, doi: 10.1016/j.knosys.2020.105887.
[25] C. Xiang, D. L. Wang, Y. Pan, A. R. Chen, X. Y. Zhou, and Y. Q. Zhang, "Accelerated topology optimization design of 3D structures based on deep learning," Structural and Multidisciplinary Optimization, vol. 65, no. 3, Mar 2022, Art no. 99, doi: 10.1007/s00158-022-03194-0.
[26] N. Napier, S. A. Sriraman, H. T. Tran, and K. A. James, "An Artificial Neural Network Approach for Generating High Resolution Designs From Low Resolution Input in Topology Optimization," Journal of Mechanical Design, vol. 142, no. 1, Jan 2020, Art no. 011402, doi: 10.1115/1.4044332.
[27] J. Shin and C. Kim, "Bi-directional evolutionary 3D topology optimization with a deep neural network," Journal of Mechanical Science and Technology, vol. 36, no. 7, pp. 3509-3519, Jul 2022, doi: 10.1007/s12206-022-0628-2.
[28] V. N. Hoang, N. L. Nguyen, D. Q. Tran, Q. V. Vu, and H. Nguyen-Xuan, "Data-driven geometry-based topology optimization," Structural and Multidisciplinary Optimization, vol. 65, no. 2, Feb 2022, Art no. 69, doi: 10.1007/s00158-022-03170-8.
[29] S. Zheng, Z. Z. He, and H. L. Liu, "Generating three-dimensional structural topologies via a U-Net convolutional neural network," Thin-Walled Structures, vol. 159, Feb 2021, Art no. 107263, doi: 10.1016/j.tws.2020.107263.
[30] J. Yan et al., "Deep learning driven real time topology optimisation based on initial stress learning," Advanced Engineering Informatics, vol. 51, Jan 2022, Art no. 101472, doi: 10.1016/j.aei.2021.101472.
[31] Y. A. Xu et al., "Machine learning based topology optimization of fiber orientation for variable stiffness composite structures," International Journal for Numerical Methods in Engineering, vol. 122, no. 22, pp. 6736-6755, Nov 2021, doi: 10.1002/nme.6809.
[32] D. W. Abueidda, S. Koric, and N. A. Sobh, "Topology optimization of 2D structures with nonlinearities using deep learning," Computers & Structures, vol. 237, Sep 2020, Art no. 106283, doi: 10.1016/j.compstruc.2020.106283.
[33] C. Qian and W. J. Ye, "Accelerating gradient-based topology optimization design with dual-model artificial neural networks," Structural and Multidisciplinary Optimization, vol. 63, no. 4, pp. 1687-1707, Apr 2021, doi: 10.1007/s00158-020-02770-6.
[34] J. Rade et al., "Algorithmically-consistent deep learning frameworks for structural topology optimization," Engineering Applications of Artificial Intelligence, vol. 106, Nov 2021, Art no. 104483, doi: 10.1016/j.engappai.2021.104483.
[35] L. Xue, J. Liu, G. L. Wen, and H. X. Wang, "Efficient, high-resolution topology optimization method based on convolutional neural networks," Frontiers of Mechanical Engineering, vol. 16, no. 1, pp. 80-96, Mar 2021, doi: 10.1007/s11465-020-0614-2.
[36] S. Doi, H. Sasaki, and H. Igarashi, "Multi-Objective Topology Optimization of Rotating Machines Using Deep Learning," Ieee Transactions on Magnetics, vol. 55, no. 6, Jun 2019, Art no. 7202605, doi: 10.1109/tmag.2019.2899934.
[37] C. Yang, Y. Kim, S. Ryu, and G. X. Gu, "Using convolutional neural networks to predict composite properties beyond the elastic limit," Mrs Communications, vol. 9, no. 2, pp. 609-617, Jun 2019, doi: 10.1557/mrc.2019.49.
[38] J. Asanuma, S. Doi, and H. Igarashi, "Transfer Learning Through Deep Learning: Application to Topology Optimization of Electric Motor," Ieee Transactions on Magnetics, vol. 56, no. 3, Mar 2020, Art no. 7512404, doi: 10.1109/tmag.2019.2956849.
[39] G. Kazakis and N. D. Lagaros, "A Simple Matlab Code for Material Design Optimization Using Reduced Order Models," Materials, vol. 15, no. 14, Jul 2022, Art no. 4972, doi: 10.3390/ma15144972.
[40] C. Y. Deng, Y. Z. Wang, C. Qin, Y. Fu, and W. Lu, "Self-directed online machine learning for topology optimization," Nature Communications, vol. 13, no. 1, Jan 2022, Art no. 388, doi: 10.1038/s41467-021-27713-7.
[41] T. Hsu et al., "Microstructure Generation via Generative Adversarial Network for Heterogeneous, Topologically Complex 3D Materials," Jom, vol. 73, no. 1, pp. 90-102, Jan 2021, doi: 10.1007/s11837-020-04484-y.
[42] A. Chandrasekhar and K. Suresh, "TOuNN: Topology Optimization using Neural Networks," Structural and Multidisciplinary Optimization, vol. 63, no. 3, pp. 1135-1149, Mar 2021, doi: 10.1007/s00158-020-02748-4.
[43] N. A. Kallioras, G. Kazakis, and N. D. Lagaros, "Accelerated topology optimization by means of deep learning," Structural and Multidisciplinary Optimization, vol. 62, no. 3, pp. 1185-1212, Sep 2020, doi: 10.1007/s00158-020-02545-z.
[44] N. A. Kallioras and N. D. Lagaros, "DL-SCALE: a novel deep learning-based model order upscaling scheme for solving topology optimization problems," Neural Computing & Applications, vol. 33, no. 12, pp. 7125-7144, Jun 2021, doi: 10.1007/s00521-020-05480-8.
[45] N. A. Kallioras, A. N. Nordas, and N. D. Lagaros, "Deep Learning-Based Accuracy Upgrade of Reduced Order Models in Topology Optimization," Applied Sciences-Basel, vol. 11, no. 24, Dec 2021, Art no. 12005, doi: 10.3390/app112412005.
[46] Y. Joo, Y. Yu, and I. G. Jang, "Unit Module-Based Convergence Acceleration for Topology Optimization Using the Spatiotemporal Deep Neural Network," Ieee Access, vol. 9, pp. 149766-149779, 2021, doi: 10.1109/access.2021.3125014.
[47] D. Bielecki, D. Patel, R. Rai, and G. F. Dargush, "Multi-stage deep neural network accelerated topology optimization," Structural and Multidisciplinary Optimization, vol. 64, no. 6, pp. 3473-3487, Dec 2021, doi: 10.1007/s00158-021-03028-5.
[48] D. Munoz, E. Nadal, J. Albelda, F. Chinesta, and J. J. Rodenas, "Allying topology and shape optimization through machine learning algorithms," Finite Elements in Analysis and Design, vol. 204, Jul 2022, Art no. 103719, doi: 10.1016/j.finel.2021.103719.
[49] V. Keshavarzzadeh, R. M. Kirby, and A. Narayan, "Robust topology optimization with low rank approximation using artificial neural networks," Computational Mechanics, Article vol. 68, no. 6, pp. 1297-1323, 2021, doi: 10.1007/s00466-021-02069-3.
[50] S. Choudhury et al., "Design of patient specific basal dental implant using Finite Element method and Artificial Neural Network technique," Proceedings of the Institution of Mechanical Engineers Part H-Journal of Engineering in Medicine, doi: 10.1177/09544119221114729.
[51] E. Ford, K. Maneparambil, S. Rajan, and N. Neithalath, "Machine learning-based accelerated property prediction of two-phase materials using microstructural descriptors and finite element analysis," Computational Materials Science, vol. 191, Apr 2021, Art no. 110328, doi: 10.1016/j.commatsci.2021.110328.
[52] L. Liang, M. L. Liu, C. Martin, and W. Sun, "A deep learning approach to estimate stress distribution: a fast and accurate surrogate of finite-element analysis," Journal of the Royal Society Interface, vol. 15, no. 138, Jan 2018, Art no. 20170844, doi: 10.1098/rsif.2017.0844.
[53] T. C. Spruegel, S. Bickel, B. Schleich, and S. Wartzack, "Approach and application to transfer heterogeneous simulation data from finite element analysis to neural networks," Journal of Computational Design and Engineering, vol. 8, no. 1, pp. 298-315, Feb 2021, doi: 10.1093/jcde/qwaa079.
[54] C. J. Zhong, R. Q. Feng, and Z. J. Zhang, "Prediction of Ultimate Bearing Capacity and Structural Optimization of Aluminum Alloy Plate Joints Based on Artificial Neural Network," International Journal of Steel Structures, vol. 21, no. 5, pp. 1759-1774, Oct 2021, doi: 10.1007/s13296-021-00533-7.
[55] H. P. Yao, Y. Gao, and Y. M. Liu, "FEA-Net: A physics-guided data-driven model for efficient mechanical response prediction," Computer Methods in Applied Mechanics and Engineering, vol. 363, May 2020, Art no. 112892, doi: 10.1016/j.cma.2020.112892.
[56] A. Madani, A. Bakhaty, J. Kim, Y. Mubarak, and M. R. K. Mofrad, "Bridging Finite Element and Machine Learning Modeling: Stress Prediction of Arterial Walls in Atherosclerosis," Journal of Biomechanical Engineering-Transactions of the Asme, vol. 141, no. 8, Aug 2019, Art no. 084502, doi: 10.1115/1.4043290.
[57] G. M. Zhang, Y. H. Zhang, X. M. Fu, and X. N. Zhang, "Effects of Daily Mastication on Bone Remodeling With Implant-Tooth-Supported Fixed Partial Prosthesis: A Finite Element Study," Ieee Access, vol. 7, pp. 33851-33858, 2019, doi: 10.1109/access.2019.2899854.
[58] P. Khanolkar, S. Basu, and C. McComb, "Image-based data on strain fields of microstructures with porosity defects," Data in Brief, vol. 34, Feb 2021, Art no. 106627, doi: 10.1016/j.dib.2020.106627.
[59] C. Yang, Y. Kim, S. Ryu, and G. X. Gu, "Prediction of composite microstructure stress-strain curves using convolutional neural networks," Materials & Design, vol. 189, Apr 2020, Art no. 108509, doi: 10.1016/j.matdes.2020.108509.
[60] C. Hoerig, J. Ghaboussi, and M. F. Insana, "Data-Driven Elasticity Imaging Using Cartesian Neural Network Constitutive Models and the Autoprogressive Method," Ieee Transactions on Medical Imaging, vol. 38, no. 5, pp. 1150-1160, May 2019, doi: 10.1109/tmi.2018.2879495.
[61] S. Chanda, S. Gupta, and D. K. Pratihar, "A combined neural network and genetic algorithm based approach for optimally designed femoral implant having improved primary stability," Applied Soft Computing, vol. 38, pp. 296-307, Jan 2016, doi: 10.1016/j.asoc.2015.10.020.
[62] S. Meo, A. Zohoori, and A. Vahedi, "Optimal design of permanent magnet flux switching generator for wind applications via artificial neural network and multi-objective particle swarm optimization hybrid approach," Energy Conversion and Management, vol. 110, pp. 230-239, Feb 2016, doi: 10.1016/j.enconman.2015.11.062.
[63] S. Roy, S. Dey, N. Khutia, A. R. Chowdhury, and S. Datta, "Design of patient specific dental implant using FE analysis and computational intelligence techniques," Applied Soft Computing, vol. 65, pp. 272-279, Apr 2018, doi: 10.1016/j.asoc.2018.01.025.
[64] F. B. da Silva, L. L. Corso, and C. A. Costa, "Optimization of pedicle screw position using finite element method and neural networks," Journal of the Brazilian Society of Mechanical Sciences and Engineering, vol. 43, no. 3, Feb 2021, Art no. 164, doi: 10.1007/s40430-021-02880-2.
[65] 李祈緯, "結合演化演算法與拓樸最佳化於 All-on-4 全口速定植牙之贋復設計," 國立成功大學, 國立成功大學機械工程學系學位論文, 2021.
[66] H. D. Shen, P. Di, J. Luo, and Y. Lin, "Clinical assessment of implant-supported full-arch immediate prostheses over 6 months of function," Clin. Implant Dent. Relat. Res., vol. 21, no. 3, pp. 473-481, Jun 2019, doi: 10.1111/cid.12784.
[67] X. Huang and Y. M. Xie, "Bi-directional evolutionary topology optimization of continuum structures with one or multiple materials," Computational Mechanics, vol. 43, pp. 393-401, 2009.
[68] C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, "Rethinking the inception architecture for computer vision," in Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 2818-2826.
[69] Y. Chen, H. Jiang, C. Li, X. Jia, and P. Ghamisi, "Deep Feature Extraction and Classification of Hyperspectral Images Based on Convolutional Neural Networks," IEEE Transactions on Geoscience and Remote Sensing, vol. 54, no. 10, pp. 6232-6251, 2016, doi: 10.1109/TGRS.2016.2584107.
[70] R. Durkan, P. Oyar, and G. Deste, "Effects of Cantilever Length and Implant Inclination on he Stress Distribution of Mandibular Prosthetic Restorations Constructed from Monolithic Zirconia Ceramic," International Journal of Oral & Maxillofacial Implants, vol. 35, no. 1, pp. 121-129, Jan-Feb 2020, doi: 10.11607/jomi.7744.