Uitdehaag Stefan, Wiarda Wim, Donders Timme, Kuiper Irene
Netherlands Forensic Institute, Laan van Ypenburg 6, 2497 GB, The Hague, The Netherlands.
Department of Physical Geography, Utrecht University, Heidelberglaan 2, 3584 CS, Utrecht, The Netherlands.
J Forensic Sci. 2017 Jul;62(4):861-868. doi: 10.1111/1556-4029.13313. Epub 2016 Dec 1.
Soil can play an important role in forensic cases in linking suspects or objects to a crime scene by comparing samples from the crime scene with samples derived from items. This study uses an adapted ED-XRF analysis (sieving instead of grinding to prevent destruction of microfossils) to produce elemental composition data of 20 elements. Different data processing techniques and statistical distances were evaluated using data from 50 samples and the log-LR cost (C ). The best performing combination, Canberra distance, relative data, and square root values, is used to construct a discriminative model. Examples of the spatial resolution of the method in crime scenes are shown for three locations, and sampling strategy is discussed. Twelve test cases were analyzed, and results showed that the method is applicable. The study shows how the combination of an analysis technique, a database, and a discriminative model can be used to compare multiple soil samples quickly.
通过将犯罪现场的样本与从物品上获取的样本进行比较，土壤在法医案件中可以发挥重要作用，将嫌疑人或物品与犯罪现场联系起来。本研究采用一种改良的能量色散X射线荧光光谱分析方法（筛分而非研磨，以防止微化石被破坏）来生成20种元素的元素组成数据。使用来自50个样本的数据以及对数似然比代价（C）评估了不同的数据处理技术和统计距离。性能最佳的组合，即堪培拉距离、相对数据和平方根值，被用于构建判别模型。展示了该方法在三个犯罪现场位置的空间分辨率示例，并讨论了采样策略。分析了12个测试案例，结果表明该方法是适用的。该研究展示了分析技术、数据库和判别模型的组合如何能够快速用于比较多个土壤样本。
