運用統計骨形狀模型以預測犬膝關節韌帶附著區域

Abstract

了解犬膝關節韌帶附著位置對於關節的生物力學建模和探索韌帶的活體生物力學特徵相當重要。對於前十字韌帶受損而需進行患者特異性解剖修復的手術時，準確得知前十字韌帶的解剖附著位置為手術的關鍵步驟。本研究旨在建立含有犬隻十字韌帶和副韌帶的附著輪廓之股骨和脛骨的可變形狀模板 (DST)，以用於預測犬隻個體之解剖韌帶附著區域。

本實驗收集54個CT影像以重組股骨和脛骨骨表面模型後，建構出統計骨骼形狀模型 (SSM)。而膝關節韌帶信息則是由31個後肢大體的CT影像收集而來，其中大體的韌帶附著位置以不透射線的顏料來標記。在進行形狀轉換程序以將SSM最佳匹配至個體骨模型後，每個樣本的韌帶附著輪廓被映射至變形後的SSM表面，我們即可藉此分析每個韌帶附著中心的離散度。平均的韌帶附著輪廓結合SSM統稱為DST，可用於估計個體韌帶附著區域與位置。本研究以留一交叉驗證法來評估預測的特定個體韌帶附著位置的準確性，首先排除其中一個樣本，並將剩餘的30個樣本建立新的DST後，進行形狀轉換程序來匹配排除樣本的骨骼表面形狀，並藉由DST預測出相應的韌帶附著輪廓。實驗結果得出股骨和脛骨端的估計韌帶中心和真實韌帶中心之間的歐幾里德距離其平均值 ± 標準差分別小於 1.71 ± 0.78 毫米和 2.14 ± 1.44 毫米。而將結果數據標準化後與先前人醫研究相比較，除了股骨端副韌帶附著位置外，大多數韌帶附著位置之估計準確性與人醫文獻結果相當。本實驗的標準化此研究結果表明使用DST結合形狀轉換程序可能是估計活體患者特異性膝關節韌帶附著位置的一種可行替代方法。

Knowledge regarding the ligament footprints in the canine stifle is essential for biomechanical modeling of the joint and exploration of the mechanical characteristics of the ligaments in vivo. In the clinical aspect, accurate footprint position of the cranial cruciate ligament (CrCL) is crucial for surgical planning of the patient-specific, anatomical repair for CrCL deficiency. The present study aimed to establish deformable shape templates (DSTs) of the femur and tibia in which the footprints of the cruciate and collateral ligaments are embedded. The DST was then used to predict the positions of subject-specific ligament footprints.

The statistical shape models (SSMs) of the femur and tibia were generated from 54 CT-derived bone surface models, and the information on the ligaments of the stifle joint was obtained from CT scans of 31 hindlimb specimens with radio-opaque markings on the ligament footprints. The contours of ligament footprints were mapped to the surfaces of deformed SSMs and the dispersions of ligament centroids were analyzed. The averaged footprint contours together with the SSM are referred to as the DST used for estimating ligament footprint locations. The subject-specific ligament footprints were then estimated in a leave-one-out cross-validation framework, in which the DST built upon 30 out of 31 samples was transformed to match the shape of the bone surface of the left-out specimen and thereafter generate the corresponding footprint contours. The results showed that the means ± standard deviations of the estimated errors in the femur and tibia were less than 1.71 ± 0.78 mm and 2.14 ± 1.44 mm, respectively. Normalization of the errors showed that most of the ligament footprints, except for the femoral collateral ligament footprints, had an accuracy comparable to those reported in the human study. This indicates that the DST with the presented transformation method can be an alternative approach for estimating the patient-specific footprint locations of the stifle ligaments in vivo.