PIXE based, Machine-Learning (PIXEL) supported workflow for glass fragments classification

• Published in: Talanta (Oxford) 2021-11, Vol.234, p.122608-122608, Article 122608
• Main Authors: Kaspi, Omer, Girshevitz, Olga, Senderowitz, Hanoch
• Format: Article
• Language: English
• Subjects: Forensic; Forensoinformatics; Glass fragments; Machine Learning; PIXE; Random forest
• ISSN: 0039-9140; 1873-3573
• DOI: 10.1016/j.talanta.2021.122608

This paper presents a structured workflow for glass fragment analysis based on a combination of Elemental Analysis using PIXE and Machine Learning tools, with the ultimate goal of standardizing and helping forensic efforts. The proposed workflow was implemented on glass fragments received from the Israeli DIFS (Israeli Police Force's Division of Identification and Forensic Sciences) that were collected from various vehicles, including glass fragments from different manufacturers and years of production. We demonstrate that this workflow can produce models with high (>80%) accuracy in identifying glass fragment's origins and provide a test-case demonstrating how the model can be applied in real-life forensic events. We provide a standard, reproducible methodology that can be used in many forensic domains beyond glass fragments, for example, Gun Shot Residue, flammable liquids, illegal substances, and more. [Display omitted] •A PIXE/machine learning-based model for classifying glass fragments was developed.•The workflow achieved >80% accuracy in classifying glass fragments to different car manufacturers.•The workflow performed well in a real world test case emulating a hit-and-run car accident.•Suggestions for workflow improvements were put forth.•The workflow could be extended to other problems of forensic relevance.