| 講演抄録/キーワード |
| 講演名 |
2018-07-24 13:55
Machine learning for estimating implanted knee functions using a CT-free navigation ○Belayat Hossain(UHyogo)・Takatoshi Morooka・Makiko Okuno(Hyogo C. Medicine)・Manabu Nii(UHyogo)・Shinichi Yoshiya(Hyogo C. Medicine)・Syoji Kobashi(UHyogo) MI2018-26 |
| 抄録 |
(和) |
In total knee arthroplasty (TKA), the damaged knee joint is replaced by artificial prosthesis. Patient-specific TKA surgical planning require evaluation of prosthesis because outcome of the TKA strongly depends on types of prosthesis and surgical methods, and it also differs from subject to subject. Machine learning (ML) techniques could be used to predict postoperative knee kinematics by utilizing a set of pairs of the clinical pre- and postoperative data. This study finds out the feasibility of the support vector regression (SVR) for clinical study, especially in Orthopaedics, and then its performance is compared to other ML method such as neural network (NN), generalized linear regression (GLR) to find the best ML method. It was found that the model?s prediction performance slightly differs from other ML methods. Therefore, this study recommends choosing the best ML methods (GLM and NN) with high accuracy for predictive model construction for predicting TKA outcome. |
| (英) |
In total knee arthroplasty (TKA), the damaged knee joint is replaced by artificial prosthesis. Patient-specific TKA surgical planning require evaluation of prosthesis because outcome of the TKA strongly depends on types of prosthesis and surgical methods, and it also differs from subject to subject. Machine learning (ML) techniques could be used to predict postoperative knee kinematics by utilizing a set of pairs of the clinical pre- and postoperative data. This study finds out the feasibility of the support vector regression (SVR) for clinical study, especially in Orthopaedics, and then its performance is compared to other ML method such as neural network (NN), generalized linear regression (GLR) to find the best ML method. It was found that the model?s prediction performance slightly differs from other ML methods. Therefore, this study recommends choosing the best ML methods (GLM and NN) with high accuracy for predictive model construction for predicting TKA outcome. |
| キーワード |
(和) |
Knee Implantation / Total knee arthroplasty / Kinematics / Machine learning / Predictive model / / / |
| (英) |
Knee Implantation / Total knee arthroplasty / Kinematics / Machine learning / Predictive model / / / |
| 文献情報 |
信学技報, vol. 118, no. 150, MI2018-26, pp. 21-24, 2018年7月. |
| 資料番号 |
MI2018-26 |
| 発行日 |
2018-07-17 (MI) |
| ISSN |
Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
| PDFダウンロード |
MI2018-26 |