QUANTITATIVE EVALUATION METHOD for TIBIOFIBULAR JOINT ALIGNMENT in ANKLE OSTEOARTHRITIS BASED on 3D BONE SIZE

Abstract

Purpose: A remedy for ankle osteoarthritis is artificial joint replacement surgery. By estimating the relative positional difference (alignment) between the bone before and after deformation, precise artificial joint replacement surgery can be performed. By using the estimated alignment, an artificial ankle joint can ensure high satisfaction with less pain and can perform necessary functions. Although bone alignment is currently estimated from X-rays and computed tomography (CT) images, it is difficult to measure three-dimensional (3D) data from two-dimensional (2D) images. Although 3D data can be estimated using a 2D-3D registration method, it requires capturing the bone images multiple times and imposing considerable burden on the patient. In this paper, we propose a method to estimate the 3D bone alignment based on the size of the bone using principal component analysis (PCA), which requires scanning only one data. Method: In the proposed method, PCA is used to create a 3D bone model based on the bone thickness, and the bone alignment is estimated by the Go-ICP algorithm. In this study, the images of the foot were captured using a CT device, and the 3D bone model was created by stacking CT images. For improving the accuracy of the alignment, a reference model based on the bone thickness was created using PCA. Subsequently, the bone was overlapped and the alignment was estimated. Result: For the evaluation of the accuracy of alignment in the proposed method, three types of methods were used to create a bone model. The proposed method was found to be the most accurate with respect to the true value in five of the six evaluation criteria. In addition, the alignment of the tibiofibular joint was evaluated using the proposed method. Regarding the displacement in the Z-axis direction, there were significant differences in both Stage 3B and Stage 4 compared with the healthy subjects. In addition, the magnitude of rotation in the Z-axis direction showed a significant difference in stage 3B as compared with a healthy subject. Conclusion: Using the proposed method, we have shown that bone alignment can be estimated three-dimensionally by scanning the bone data once only. In addition, by comparing with the reference model using PCA based on the thickness of the bone, the accuracy of alignment is improved as compared with the reference model, which is not based on the bone thickness.