Exploring radiomic features of lateral cerebral ventricles in postmortem CT for postmortem interval estimation

Area of Science:

• Forensic Medicine
• Radiology
• Medical Imaging

Background:

• Estimating postmortem interval (PMI) is crucial in forensic medicine.
• Traditional methods for PMI estimation have limitations.
• Quantitative imaging offers potential for improved accuracy.

Purpose of the Study:

• To investigate radiomic features from postmortem CT scans of lateral cerebral ventricles (LCVs) for PMI estimation.
• To assess the association between radiomic features and PMI using sequential scan data.
• To explore the potential of quantitative imaging in forensic investigations.

Main Methods:

• Acquired sequential postmortem computed tomography (PMCT) scans from twelve corpses (5.5–273 hours postmortem).
• Extracted radiomic features from the LCVs.
• Applied a mixed-effect model to analyze the relationship between radiomic features and PMI.

Main Results:

• Significant associations were found between several radiomic features and PMI.
• Median Hounsfield Units (HU) within LCVs showed the strongest association (p < 9.47 × 10⁻⁹).
• Other significant features included LCV surface area, L-major axis, L-minor axis, and HU entropy.
• A logarithmic model best described PMI-related changes in LCV volume and mean HU intensity.

Conclusions:

• PMCT-based radiomics can provide valuable information for PMI estimation.
• Radiomic features of LCVs show significant correlation with postmortem interval.
• Quantitative imaging using PMCT is a promising complementary tool for forensic investigations.