Findings and methodology are reported for a research project with the goal of determining the accuracy and reliability of the LatentSleuth technology and whether integrating LatentSleuth into current comparison workflow for complex comparisons improves efficiency and reproducibility compared to existing methods.
This report indicates the LatentSleuth software provided accurate results across all quality levels of latent fingerprints. Of the 600 automated searches conducted within this validation study, accurate results (the true-mate reference image located within the top five positions and on the correct corresponding ridge detail) were produced in 571 (95.2 percent) searches, with the true-mate reference image ranked in the number one position in 533 of the 600 (88.8 percent) searches conducted. The LatentSleuth software was less effective with low-quality latent prints in high comparison complexity. The software was validated for use in casework. By applying the formula achieved from the scatter graph, the benefits of an examiner incorporating LatentSleuth into comparisons is substantial in instances where examiner time is greater than 2 hours. Overall, the data do not support a single type of comparison case in which LatentSleuth is most effective or efficient. Apparently, LatentSleuth is most helpful when there are a large number of latent prints in a case, specifically lower quality prints, suggesting a remedy for examiner fatigue. Cases that involved 10 comparisons were the most common in the study. Project design, methods, and data analysis are described. 7 figures, 4 tables, and 2 references