Inter-laboratory workflow for forensic applications: Classification of car glass fragments

Area of Science:

• Forensic Science
• Nuclear Analytical Techniques
• Materials Science

Background:

• The International Atomic Energy Agency (IAEA) coordinated research to apply nuclear techniques in forensics.
• Particle Induced X-ray Emission (PIXE) is a key Ion Beam Analysis (IBA) technique for elemental analysis.

Purpose of the Study:

• To analyze and classify forensic glass specimens using PIXE.
• To develop machine learning models for determining the origin of glass samples.

Main Methods:

• PIXE measurements were performed on car window glass fragments from various manufacturers.
• Elemental compositions were analyzed using machine learning algorithms.
• Data from multiple laboratories were combined after pre-processing.

Main Results:

• Machine learning models achieved over 80% accuracy in classifying glass fragments by car model.
• A unified database combining results from different labs yielded comparable or superior classification performance.
• The methodology demonstrates potential for establishing an international forensic glass database.

Conclusions:

• PIXE combined with machine learning is effective for forensic glass analysis and classification.
• Data harmonization across laboratories enables robust and generalizable classification models.
• This approach can support law enforcement agencies globally by providing reliable forensic evidence.