A Review on Automatic Personal Identification Using Panoramic Radiographs and Computed Tomography.

Area of Science:

• Forensic Radiology
• Medical Imaging Analysis
• Biometrics

Background:

• Identifying unknown individuals is a critical challenge in forensic and emergency medicine.
• Radiological images offer unique anatomical features for personal identification.
• Current applications of radiology-based identification are largely experimental.

Purpose of the Study:

• To review current methods for automatic radiology-based personal identification.
• To evaluate the performance of these identification methods.
• To highlight potential applications in forensic and clinical settings.

Main Methods:

• A narrative review of studies published from 2018 onwards was conducted.
• PubMed and Google Scholar were used for literature search.
• Studies involving automated or semi-automated identification using panoramic radiographs (PR) or computed tomography (CT) were included.

Main Results:

• 32 studies were included, focusing on PR-to-PR, CT-to-CT, and CT-to-PR identification.
• Descriptor-based computer vision was the most common approach (9 studies).
• Deep learning methods showed promise for feature extraction, classification, and bone segmentation.

Conclusions:

• Descriptor-based computer vision is flexible and robust for large-database comparisons and postmortem identification.
• Deep learning approaches require further validation for cross-individual matching.
• Standardized databases and automated pipelines are crucial for advancing radiology-based identification.