Quantitative Analysis of Lower Limb Alignments for Joint Disease Patient Based on Template Registration

Numerous studies show that accurate reconstruction of lower limb alignments is one of the main determinants of the long-term success of Total knee arthroplasty (TKA). The purpose of this paper is to explore a new 3D CT measurement method base on template registration to extract the lower limb alignments quickly and accurately. By constructing statistical shape model of leg and using non-rigid template matching method, we can accurately extract key anatomical structure feature points of lower limb. A bidirectional-section method is proposed to locate the femoral head center. Then, the feature points are connected to obtain the femoral mechanical axis and the tibial mechanical axis. The proposed method is verified in 20 cases of CT data. The experimental results demonstrate that the average of the feature point extraction error is less than 3 mm, the average angle error between the mechanical axes and the ground-truth is less than 0.6°. Compared with the benchmark method, the measurement accuracy of the feature points is improved by 39.75%, the processing speed increased by 37.50%. The proposed method can extract lower limb alignments accurately, it will help the long-term success of TKA surgery.