Milheiro A, De Tobel J, Capitaneanu C, Shaheen E, Fieuws S, Thevissen P
Department of Imaging and Pathology, Forensic Odontology, KU Leuven, Louvain, Belgium.
Diagnostic Sciences - Radiology, Ghent University, Corneel Heymanslaan 10, 9000, Ghent, Belgium.
Int J Legal Med. 2024 Jan;138(1):25-34. doi: 10.1007/s00414-022-02853-7. Epub 2022 Jun 15.
In forensic identification, lack of eccentric characteristics of intact dentitions hinders correct ante-mortem/post-mortem (AM/PM) matching. It remains unclear which morphological dental parameters hold strong potential as identifiers. This study aimed to establish a method to quantify and rank the identifying potential of one (or a combination of) continuous morphological parameter(s), and to provide a proof of concept. First, a statistic was defined that quantifies the identifying potential: the mean potential set (MPS). The MPS is derived from inter-observer agreement data and it indicates the percentage of subjects in the AM reference dataset who at least need to be considered to detect the correct PM subject. This was calculated in a univariate and a multivariate setting. Second, the method was validated on maxillary first molar crowns of 82 3D-digitally scanned cast models. Standardized measurements were registered using 3D modeling software (3-Matic Medical 12.0, Materialise N.V., Leuven, Belgium): tooth depth, angles between cusps, distances between cusps, distances between the cusps, and the mesial pit. A random sample of 40 first molars was measured by a second examiner. Quantifying and ranking the parameters allowed selecting those with the strongest identifying potential. This was found for the tooth depth (1 measurement, MPS = 17.1%, ICC = 0.879) in the univariate setting, and the angles between cusps (4 measurements, MPS = 3.9%) in the multivariate setting. As expected, the multivariate approach held significantly stronger identifying potential, but more measurements were needed (i.e., more time-consuming). Our method allows quantifying and ranking the potential of dental morphological parameters as identifiers using a clear-cut statistic.
在法医鉴定中，完整牙列缺乏独特特征阻碍了生前/死后（AM/PM）的正确匹配。目前尚不清楚哪些牙齿形态参数具有很强的识别潜力。本研究旨在建立一种方法，对一个（或一组）连续形态参数的识别潜力进行量化和排序，并提供概念验证。首先，定义了一个量化识别潜力的统计量：平均潜力集（MPS）。MPS源自观察者间一致性数据，它表示在AM参考数据集中至少需要考虑的受试者百分比，以检测出正确的PM受试者。这在单变量和多变量设置中进行了计算。其次，该方法在82个三维数字扫描铸模模型的上颌第一磨牙牙冠上进行了验证。使用三维建模软件（3-Matic Medical 12.0，Materialise N.V.，比利时鲁汶）记录标准化测量值：牙深度、牙尖之间的角度、牙尖之间的距离、牙尖之间的距离以及近中窝。随机抽取40颗第一磨牙由另一位检查者进行测量。对参数进行量化和排序可以选择具有最强识别潜力的参数。在单变量设置中发现牙深度（1次测量，MPS = 17.1%，ICC = 0.879）具有最强识别潜力，在多变量设置中发现牙尖之间的角度（4次测量，MPS = 3.9%）具有最强识别潜力。正如预期的那样，多变量方法具有明显更强的识别潜力，但需要更多测量（即更耗时）。我们的方法允许使用明确的统计量对牙齿形态参数作为标识符的潜力进行量化和排序。
