论文部分内容阅读
目的选择骨盆CT片上的特征指标,建立逐步回归方程,探讨其在法医学同一认定中的应用价值。方法收集160名不同被检查者骨盆CT影像片各1张,70名被检查者不同次骨盆CT影像片各2张。选择并测量骨盆CT片上的14项指标值,分别计算不同人随机分组相同测量指标的组间的差值,以及相同人不同次测量指标间的差值,运用二分类logistic逐步回归分析,建立各项指标的一元回归方程和多项指标的多元回归方程,并对方程进行盲测检验。结果建立的14个一元方程中同一认定的正确率在61.1%(骶骨耳状面后缘宽)~80.5%(第一骶椎平面左右髂骨前端间距)之间;建立的6个多元回归方程的正确率在80.5%~93.8%之间。盲测准确率为100%。结论本文在CT片上选择的14项特征指标可以用于同一认定,在使用时应尽可能选用多元指标以得到更准确的结果。
Objective To select the characteristic indexes of pelvic CT and establish the stepwise regression equation to explore its application value in the same identification of forensic medicine. Methods One pelvis CT image of 160 subjects was collected, and the other 2 pelvic CT images of 70 subjects under examination were collected. 14 indexes on pelvic CT were selected and measured, and the difference between groups randomly assigned to the same measurement index and the difference between the same person in different measurement indexes were calculated. Two-step logistic stepwise regression analysis was used to establish The regression equation of one item and the multiple regression equation of multiple indicators, and the equation is blindly tested. Results The correctness of the same identification in the 14 univariate equations was between 61.1% (posterior margin of the sacral auricular surface) to 80.5% (the distance between the iliac anterior ends of the first sacral plane); the six multiple regression equations established The correct rate of 80.5% to 93.8%. Blind test accuracy of 100%. Conclusion The 14 feature indexes selected in this paper can be used for the same identification. When used, multivariate indexes should be used as far as possible to get more accurate results.