以影像評估藍瑞斯與約克夏女豬之腿部結構與強健度

Abstract

種母豬的腿部強健度會影響其使用年限，而且跛足或腿部問題是生產母豬被淘汰的第二大原因，僅次於繁殖障礙，特別是前三胎的母豬。由於腿部結構相關性狀為中低遺傳率，因此在豬隻生長過程中觀察腿部結構是選種時必要的工作之一。近年已有一些研究致力於開發量化的檢測方法以評估豬隻跛行，然而尚未有長期追蹤新女豬腿部結構變化的研究，這些量化評估方法也尚未應用於現場，因此本研究將24頭新女豬在3.0、4.5、6.0、7.5月齡時，追蹤前、後肢的關節夾角及步態資料對四肢腿部受力(Ground reaction force, GRF)建立腿部受力的預測公式。 使用運動學步態分析關節角度資料，其中，前肢肩關節角度在3.0月齡時顯著小於其他月齡、前肢蹄面與地面夾角在7.5月齡時顯著小於其他月齡(p<0.05)；後肢膝關節角度及後肢飛節角度在3.0月齡時顯著小於其他月齡、後肢蹄面與地面夾角在7.5月齡時顯著大於3.0月齡(p<0.05)。運用壓力墊之動力學步態分析方法取得站立時間、腿部擺動時間、步態週期、步幅等資料。為建立預測腿部受力公式，運用前(後)肢的腿部受力數據與各關節角度及步態數據進行向後消去複迴歸分析，自變數為女豬的日齡、體重、站立時間、腿部擺動時間、步態週期、步幅及前(後)肢各關節角度。前肢分析結果為日齡、站立時間、肩關節角度、肘關節角度是對前肢受力預測具顯著效應的自變數，R^2=0.75；後肢分析結果為日齡、站立時間、後膝關節角度、髖關節角度是對後肢受力預測具顯著效應的自變數，R^2=0.69。進一步以壓力墊得到的腿部受力對稱指數及預測公式之腿部受力對稱指數及步幅對稱指數帶入混淆矩陣，得到各種方法的準確度，做為腿部強健度評估的指標。 利用混淆矩陣對目視女豬淘汰結果、受力預測對稱指數結果、壓力墊受力對稱指數結果與步幅對稱指數結果進行分類，並計算準確度、敏感性及和特異性分析。現場人員目視選留結果與7.5月齡步幅對稱指數的準確度為85%、敏感性僅36%、特異性達100%，顯示目視選留僅能挑出少量之步態異常豬隻。而以複迴歸公式所得之受力預測對稱指數與壓力墊受力對稱指數做比較，得到之準確度為85%、敏感性及特異性分別為54%及94%，複迴歸分析預測公式之敏感性較現場人員目視選留更高，代表在預測公式有判別出步態異常豬隻的潛力。 本研究利用女豬選留前拍攝之步態影像計算關節角度及站立時間等數據，透過複迴歸預測公式預估腿部受力及對稱指數，未來可在沒有壓力墊設備時做為女豬選留前腿部強健度的評估指標。

The durability of sows would be affected by their leg soundness. Problems of leg or lameness is the second reason for culling, second only to reproductive disorder, especially in the first three parities. Due to the intermediate to low heritability of traits related to leg structure, it is necessary to observe the structure during the growing period of pigs in selection programs. In recent years, some studies have focused on quantitative methods for measuring pig lameness. However, a study on the long-term tracking of changes in legs and feet structure has not been conducted. In addition, those quantitative evaluation methods have not yet been applied on farms. Therefore, the objective of this study was to collect the ground reaction force (GRF) and data sets of joint angle and gait for fore- and hind-limbs of 24 purebred gilts at 3.0, 4.5, 6.0, 7.5 months of age to construct the prediction equation of GRF. In the data of joint angle analyzed by kinematic gait analysis, the shoulder joint angle of forelimb at 3.0 months of age was significantly smaller than those at other months of age. The angle between the hoof surface of the forelimb and the ground at 7.5 months of age was significantly smaller than those at other months of age (P < 0.05). The angles of knee joint and hock joint of the hindlimb at 3.0 months of age was significantly smaller than those at other months of age. The angle between the hoof of the hindlimb and the ground at 7.5 months of age was significantly larger than that at 3.0 months of age (P < 0.05). In addition, the stance time, swing time, stride time, stride length, and GRF were obtained by using gait analysis of pressure mat. To establish the prediction equations for GRF on feet, multiple regression analyses were conducted using data sets of GRF on legs and feet, joint angles, and gait of fore- and hind-limbs with a backward selection procedure. The results of prediction equation for forelimb showed significant effects on days of age, stance time, angle of shoulder joint and elbow joint on GRF with R^2=0.75. For hindlimb, the results of prediction equation have significant effects on days of age, stance time, angle of stifle joint and hip joint on force on legs and feet with R^2=0.69. The GRF symmetry index obtained from the pressure mat and the GRF prediction equation symmetry index and stride length symmetry index were brought into the confusion matrix to obtain the prediction accuracy of various methods and used them as an index for leg soundness. The confusion matrix was applied to classify visual inspection for gilt culling, force prediction symmetry index, GRF symmetry index, and stride symmetry index, with their accuracy, sensitivity, and specificity. The results of visual inspection for gilt culling vs. stride symmetry index at 7.5 months of age showed accuracy, sensitivity, and specificity of 85%, 36%, and 100%, respectively. These results indicated that visual inspection for gilt culling can only identify a small number of pigs with a symmetrical gait. However, the accuracy, sensitivity, and specificity of the force prediction equation symmetry index compared to the pressure mat force symmetry index were 85%, 54%, and 94%, respectively. The sensitivity of the multiple regression analysis prediction formula is higher than that of visual inspection, so the prediction equation has the potential to identify pigs with an asymmetry gait. This study utilized gilt gait videos to calculate joint angles with standing time data estimated by the multiple regression prediction equation to estimate the GRF of legs. In the future, these could be used as evaluation indices for the leg soundness of gilts.