Gallidabino M, Romolo F S, Weyermann C
Université de Lausanne, École des Sciences Criminelles, Bâtiment Batochime, 1015 Lausanne-Dorigny, Switzerland; King's College London, Analytical and Environmental Science Division, 150 Stamford Street, SE1 9NH London, United Kingdom.
Sapienza Università di Roma, Section of Legal Medicine, Department SAIMLAL, Viale Regina Elena 336, 00161 Rome, Italy.
Forensic Sci Int. 2017 Mar;272:171-183. doi: 10.1016/j.forsciint.2016.12.027. Epub 2016 Dec 27.
Estimating the time since discharge of spent cartridges can be a valuable tool in the forensic investigation of firearm-related crimes. To reach this aim, it was previously proposed that the decrease of volatile organic compounds released during discharge is monitored over time using non-destructive headspace extraction techniques. While promising results were obtained for large-calibre cartridges (e.g., shotgun shells), handgun calibres yielded unsatisfying results. In addition to the natural complexity of the specimen itself, these can also be attributed to some selective choices in the methods development. Thus, the present series of papers aimed to systematically evaluate the potential of headspace analysis to estimate the time since discharge of cartridges through the use of more comprehensive analytical and interpretative techniques. Following the comprehensive optimisation and validation of an exhaustive headspace sorptive extraction (HSSE) method in the first part of this work, the present paper addresses the application of chemometric tools in order to systematically evaluate the potential of applying headspace analysis to estimate the time since discharge of 9mm Geco cartridges. Several multivariate regression and pre-treatment methods were tested and compared to univariate models based on non-linear regression. Random forests (RF) and partial least squares (PLS) proceeded by pairwise log-ratios normalisation (PLR) showed the best results, and allowed to estimate time since discharge up to 48h of ageing and to differentiate recently fired from older cartridges (e.g., less than 5h compared to more than 1-2 days). The proposed multivariate approaches showed significant improvement compared to univariate models. The effects of storage conditions were also tested and results demonstrated that temperature, humidity and cartridge position should be taken into account when estimating the time since discharge.
估算已发射弹药筒的发射时间，对于涉枪犯罪的法医调查而言可能是一项重要工具。为实现这一目标，此前有人提出，利用无损顶空萃取技术，监测发射过程中释放的挥发性有机化合物随时间的减少情况。虽然大口径弹药筒（如猎枪子弹）取得了不错的结果，但手枪口径的弹药筒却不尽人意。除了样本本身固有的复杂性外，这些结果也可归因于方法开发中的一些选择性选择。因此，本系列论文旨在通过使用更全面的分析和解释技术，系统地评估顶空分析在估算弹药筒发射时间方面的潜力。在本研究第一部分对详尽的顶空吸附萃取（HSSE）方法进行全面优化和验证之后，本文探讨化学计量工具的应用，以便系统地评估应用顶空分析估算9毫米格科弹药筒发射时间的潜力。测试了几种多元回归和预处理方法，并与基于非线性回归的单变量模型进行比较。经成对对数比归一化（PLR）处理的随机森林（RF）和偏最小二乘法（PLS）显示出最佳结果，能够估算长达48小时老化时间的发射时间，并区分近期发射的弹药筒和较旧的弹药筒（例如，小于5小时与超过1 - 2天相比）。与单变量模型相比，所提出的多元方法有显著改进。还测试了储存条件的影响，结果表明，在估算发射时间时应考虑温度、湿度和弹药筒位置。
