De-Giorgio Fabio, Guerreri Michele, Gatta Roberto, Bergamin Eva, De Vita Vittorio, Mancino Matteo, Boldrini Luca, Sala Evis, Pascali Vincenzo L
Fondazione Policlinico Universitario A. Gemelli IRCCS, Rome, Italy.
Department of Healthcare Surveillance and Bioethics, Section of Legal Medicine, Università Cattolica del Sacro Cuore, Rome, Italy.
Int J Legal Med. 2025 Mar;139(2):667-677. doi: 10.1007/s00414-024-03396-9. Epub 2024 Dec 20.
The aim of this study is to investigate the potential of radiomic features extracted from postmortem computed tomography (PMCT) scans of the lateral cerebral ventricles (LCVs) to provide information on the time since death, or postmortem interval (PMI), a critical aspect of forensic medicine. Periodic PMCT scans, referred to as "sequential scans", were obtained from twelve corpses with known times of death, ranging from 5.5 to 273 h postmortem. Radiomics features were then extracted from the LCVs, and a mixed-effect model, specifically designed for sequential data, was employed to assess the association between feature values and PMI. Four model variants were fitted to the data to identify the best functional form to explain the relationship between the variables. Significant associations were observed for features, the most significant being the median Hounsfield Units (HU) within the LCVs (p < 9.47 × 10⁻⁹), LCVs surface area (p < 4.69 × 10⁻⁶), L-major axis (p < 2.17 × 10⁻⁵), L-minor axis (p < 1.30 × 10⁻⁴), and HU entropy (p < 4.16 × 10⁻⁴). Our findings align with previous studies, supporting a logarithmic model for PMI-related changes in LCV volume and mean HU intensity value. This study highlights the potential of PMCT-based radiomics as source of complementary information that could be integrated into existing methods for PMI estimation. Our results support the application of a quantitative imaging approach in forensic investigations.
本研究的目的是调查从外侧脑室（LCV）的死后计算机断层扫描（PMCT）中提取的放射组学特征，以提供关于死亡时间或死后间隔（PMI）的信息，这是法医学的一个关键方面。从12具已知死亡时间的尸体上获得了定期的PMCT扫描，即“连续扫描”，死后时间从5.5小时到273小时不等。然后从LCV中提取放射组学特征，并采用专门为连续数据设计的混合效应模型来评估特征值与PMI之间的关联。对数据拟合了四种模型变体，以确定解释变量之间关系的最佳函数形式。观察到特征之间存在显著关联，最显著的是LCV内的中位亨氏单位（HU）（p < 9.47×10⁻⁹）、LCV表面积（p < 4.69×10⁻⁶）、L长轴（p < 2.17×10⁻⁵）、L短轴（p < 1.30×10⁻⁴）和HU熵（p < 4.16×10⁻⁴）。我们的研究结果与之前的研究一致，支持LCV体积和平均HU强度值与PMI相关变化的对数模型。本研究强调了基于PMCT的放射组学作为补充信息来源的潜力，可将其整合到现有的PMI估计方法中。我们的结果支持在法医调查中应用定量成像方法。
