Analyzing clinical knee kinematics data for post-operative kinematics prediction

抄録

Knee implantation is a popular knee surgery to replace damaged knee joint in total knee arthroplasty (TKA). It is very essential to predict post-operative knee kinematic before surgery for successful surgical planning. However, the key factor is that outcome of TKA operation strongly depends on types of prosthesis and surgical methods, and also differs from subject to subject. It is well known, patient-specific knee implant and surgical method increases operation quality. Thus, the requirement of analyzing pre-operative knee kinematic for predicting post-operative knee function for patient-specific TKA surgical planning. This study aids the medical surgeon is this regard. The proposed method was constructed in 45 clinical knee data using machine learning technique, and also verified by leave-one-out-cross validation (LOOCV) test. This study mainly focuses on model performance with and without feature extraction method. Results show that first one outperforms over no-PCA based with mean correlation coefficient of 0.68, and root mean squared error of 5.84 mm.