Firearm Forensics Black-Box Studies for Examiners and Algorithms using Measured 3D Surface Topographies

• NCJ Number: 254338
• Author(s): Ryan Lilien
• Date Published: May 2019
• Length: 38 pages
• Grant Number(s): 2017-IJ-CX-0024
• Agencies: NIJ-Sponsored
• Publication Type: Research (Applied/Empirical); Report (Study/Research); Report (Grant Sponsored); Program/Project Description

Annotation

This Final Research Performance Progress Report presents the findings and methodology of a research project that took critical steps toward further validation of tool mark examination and the use of 3D scanning technology in the forensic lab.

Abstract

Specifically, the project improved the research team's Virtual Microscopy Viewer software, selected and scanned a large set of cartridge cases, and conducted a large black-box virtual microscopy study. The comparison of cartridge cases was based on the observation that microscopic firearm imperfection can be transferred to ammunition during firing. The ability to certify two cartridge cases as similar is therefore a function of both the ability to capture and visualize a high-resolution measurement of each specimen and the ability to identify and match relevant structural features between the two. Courtroom challenges and recent reports have called for additional research into underlying error rates and performance measures for these comparative methods. This report indicates that the proposed goals of the study were achieved during the project period. The project developed and updated the Virtual Comparison Microscopy (VCM) testing platform capable of supporting most modern Windows computers. The updated software has a testing mode that is designed to facilitate implementation of validation studies. An algorithm-testing software tool was developed to assist in the future evaluation of different comparison algorithms. The project also selected test fires for 40 test sets that included firearms with a range of tool-mark types and complexity, Test fires were scanned and assembled for the validation study. The project completed the largest VCM validation and error-rate study to date. It involved 107 participants that included 76 qualified examiners from the United States and Canada. The examiners achieved highly accurate analysis (0.2 percent error rate). 16 figures and 21 references