犬隻動作分析中後肢軟組織移動誤差與標記組選擇之研究

Abstract

在獸醫臨床醫學領域，評估健康犬隻、骨關節或神經性疾病患犬的運動力學時， 已廣泛使用皮膚標記點為基礎的動作分析系統。然而，此方法會受到無法預測的軟 組織移動誤差影響，導致追蹤皮膚標記點時無法真實呈現骨骼運動狀況。過去人類 動作分析研究也曾報導在同一次動作若使用不同的皮膚標記點作為追蹤目標，會 計算出不同的關節角度。目前在犬隻動作分析尚無一致的皮膚標記點模型、或是標 準黏貼位置選擇。此研究目標希望評估犬後肢皮膚標記點組合的皮膚移動誤差，與 使用不同的標記點組合造成的膝關節角度偏移狀況。 研究共招募十隻混種中大型成犬，排除骨骼肌肉異常與系統性疾病後，進行電 腦斷層掃描，取得病患股骨脛骨骨骼表面模型。接著於大小腿黏貼 19 顆紅外線反 光球，並同時以透視攝影設備與紅外線攝影機拍攝後肢被動屈曲伸展運動狀況。本 研究使用模型為基礎的影像追蹤與定位技術，透過電腦斷層取得之骨頭表面模型 與膝關節透視攝影影響，取得股骨脛骨於三維空間中的位置，並將計算出的關節角 度定為參考值。將紅外線攝影機取得之不同皮膚標記點，依據人類動作分析研究之 標記點組合設計原則，分別於股骨與脛骨選擇數組適當標記點組合作為追蹤目標。 使用普氏分析(Procrustes Analysis)量化標記點組合的形狀移動與改變，將位移量、 旋轉量、縮放量與變形量作為標記點組合的皮膚移動誤差參數。接著，搭配大小腿 不同的標記點組合計算出膝關節角度，並評估計算出的角度與參考角度差異。除此 之外，本研究也使用不同的標記點組合選擇方式，選取黏貼於骨特徵點與非骨特徵 點的皮膚標記點組成標記點組合，觀察標記點組成是否影響標記點組合之整體皮 膚移動誤差。研究結果顯示使用皮膚標記點位置計算膝關節角度，在膝關節屈曲角 度超過 50% 時，計算出之屈曲/伸展、內收/外展、內轉/外轉角度有很高比例與參 考角度有顯著差異。在屈曲/伸展動作中，因關節活動度較大，誤差造成之影響相 對較小;在內收/外展動作，不同標記點組合搭配計算之關節角度差異較大，使用 位移量與旋轉量較小的股骨標記點組合搭配位移量較小的脛骨標記點組合，可以 計算出較準確的關節角度;在內轉/外轉角度中，本研究選擇的標記點組合皆高估 脛骨於膝關節屈曲時的內轉角度。除此之外，犬隻後肢細長的特徵也影響標記點組成，使符合設計原則的標記點組合數量有限。目前在犬後肢動作分析尚無統一皮膚 標記點模型，在進行研究結果比較時建議需將此差異列入考量。未來研究在進行後 肢動作分析時，建議選用誤差較少之標記點組合進行關節角度計算。

Skin marker-based motion analysis had been widely used in veterinary medicine for quantitative investigations of kinematics in healthy and diseases dogs. However, the precision and accuracy were compromised by the unpredictable interference of soft tissues between markers and bones, namely the Soft Tissue Artifacts (STA). Previous studies in human motion analysis have demonstrated that marker clusters with different marker locations led to discrepant joint angles when describing identical joint movement. Currently, no consistent protocol or recommended cluster composition of canine gait analysis was proposed. Therefore, the objectives of this study are to evaluate quantitatively the STA in cluster level and the effects of different marker cluster compositions on the calculated kinematics variables of canine stifle joint. Ten adult mixed-breed dogs without history of musculoskeletal diseases and detectable abnormality during physical and orthopedic examination were enrolled in the IACUC approved study. Each patient underwent a CT scan to acquire a subject-specific volumetric data of the femur and tibia. Nine retroreflective markers were affixed to thigh and ten to shank at several anatomic and non-anatomic landmarks. A laboratory space was equipped with a single-plane video-fluoroscopy system to collect real-time fluoroscopic images of the tibiofemoral joint during movement. The testing space was also equipped with a motion analysis system to simultaneously track the retroreflective markers during tests. A subject calibration trial and three successful trials of isolated stifle passive extension movement were collected when the subjects were under general anesthesia. The true bone poses were determined using a validated model-based tracking method integrating real-time fluoroscopic images and volumetric CT data of bones, which were used to derive the standard reference of the stifle kinematics. The STA in cluster level was first studied to observe the different patterns in clusters composing with markers that were attached to anatomical or non-anatomical landmarks. Next, thigh and shank clusters were selected from all four-marker combinations with isotropy index greater than 0.5. Based on Procrustes superimposition approach for shape analysis, cluster-level STAs of each cluster composition were expressed in four variables: position, orientation, size and shape ranges, following previous studies. Clusters with smallest and largest position and orientation ranges were paired to calculate stifle joint angles in three anatomical planes. Paired t-tests were used to exam the difference between joint angles derived using different cluster compositions and the standard reference angles. The marker composition with smallest error in the description of joint kinematics was suggested in future research and clinical gait analysis. The findings would contribute to improve our knowledge of STA in canine hindlimb and their effect on joint kinematics measurement.