Journal of forensic sciences最新文献

Since the late 20th century, wrongful convictions based on fingerprint evidence and judicial scrutiny have raised questions about the scientific validity of fingerprint evidence, necessitating research into the scientific foundations of fingerprint identification. This article employs state-of-the-art AI algorithms to achieve fingerprint pattern classification, pose estimation, and minutiae detection. Based on a large-scale dataset of 620,211 fingerprint images, various analytical methods are applied to explore the relationships between fingerprint patterns, minutiae, and fingers, and statistical analysis is conducted on the quantity and spatial distribution of six types of minutiae: ridge endings, bifurcations, spurs, independent ridges, lakes, and crossovers. Compared with previous research, the average accuracy of minutiae detection is improved from 97.22% to 99.45%. Results indicate that whorls are the most common pattern, associated with thumbs and ring fingers, while loops are associated with middle and little fingers. The overall distribution ranges (in percentages) of the six types of minutiae are: ridge endings [58.288, 58.875], bifurcations [37.874, 38.421], spurs [1.301, 1.314], independent ridges [1.246, 1.260], lakes [0.415, 0.419], and crossovers [0.291, 0.295]. Spatial distribution analysis reveals that independent ridges exhibit concentrated distribution in delta regions, while the other types of minutiae are primarily concentrated in the core regions. This article quantifies the evidential value of different minutiae by analyzing the relationships among patterns, minutiae, and fingers as well as the spatial distribution of minutiae, providing a scientific statistical foundation for establishing probabilistic fingerprint identification models and contributing to improving objectivity and scientific rigor in fingerprint identification.

Crime scene exhibits are often packaged at a crime scene and transported to a laboratory for DNA analysis. DNA-containing material may be lost from the sampling site of the exhibit to the inside of the packaging, preventing identification of a suspect, or may transfer to other parts of the exhibit complicating the interpretation of results. We sought to mitigate this DNA transfer by testing packaging that reduced direct contact with the exhibit, limited the exhibit's movement, or contained physical barriers to separate areas of the exhibit. Blood, saliva, or touch DNA were deposited onto mock exhibits that were packaged by one of four methods: unsecured, secured to bottom, secured suspended, or secured suspended with barrier separating areas. Packaged exhibits were then transported in a manner resembling casework, after which the location and amount of DNA on the exhibit and packaging were assessed. Control samples, which were not transported, were also tested. Touch and saliva deposits appeared to transfer by direct contact with the packaging and this transfer could be mitigated by suspending and/or securing the exhibits within packaging to minimize contact. Blood flaking from the exhibits meant the transfer of blood was inevitable under the conditions tested. While limiting direct contact between the exhibit and packaging minimized relocation of blood on the exhibit, the use of physical barriers prevented its transfer to other parts of the packaging. We show that while DNA transfer in packaging is not uncommon, there are strategies to mitigate this.

This study presented a large-scale statistical examination of 168,974 tenprint records to evaluate whether pattern distribution across the fingers is random or exhibits structured interdependence, and whether sex-related differences in pattern frequency exist. Two Bayesian networks were empirically developed and validated to model the relationships between the pattern types of different fingers. The first network focused specifically on the occurrence of whorls and was evaluated relative to established frequencies of the Henry primary classification system, revealing expected relationships between pattern types, but also extending beyond, traditional classification approaches. The second network incorporated all major fingerprint pattern types to model probabilistic dependencies across fingers and hands. This work demonstrates and models significant inter- and intrahand relationships. Additionally, the developed Bayesian networks enable automated biometric identification system users to input their data to model finger variation for the computation of statistical conclusions. These relationships can be leveraged to predict pattern occurrences on other fingers which can be used to limit file penetration by filtering searches by finger position yielding increased search accuracy through a reduced search gallery.

The hypothesis of fingerprint individuality continues to be debated due to limited empirical verification, impacting the scientific foundation of fingerprint identification. This study proposed a quantitative model for fingerprint individuality and investigated the three-dimensional (3D) distribution of minutiae. This model considered the position and direction of minutiae as 3D feature variables. We extracted 3D feature data from 56,812,114 known fingerprints based on the automatic fingerprint identification system (AFIS). Following data calibration, translation, and error correction, we statistically analyzed the distribution density of minutiae. We developed the algorithm to calculate the individuality score of a single fingerprint through the individuality model. The experimental results showed that the minutiae distribution followed distinct patterns. The distribution density of minutiae exhibited symmetry between corresponding fingers on left/right hands. Significant variations in minutiae distribution density and central point distribution were observed across the five pattern types (whorl, left loop, right loop, arch, accidental). Minutiae with different directions exhibited symmetry along the Y-axis in both positional and quantitative distribution. Minutiae within diagonally opposite angular ranges showed similar distribution trends. The individuality scores were robust to distinguish different fingerprints. We preliminarily applied the individuality score to provide a basis for modifying the AFIS scoring mechanism, and we found that the individuality score of same-source fingerprints was greater than that of close nonmatches (CNMs). This work provides novel insights into fingerprint individuality and establishes a statistical foundation for refining AFIS scoring mechanisms and likelihood-ratio evidence evaluation frameworks.

Malware detection and classification in network traffic is a critical challenge in cybersecurity, with evolving threats that traditional methods struggle to address. As network traffic becomes more complex, accurately identifying malicious activities while minimizing false positives is essential for real-time monitoring systems. This study aims to enhance malware detection using deep learning (DL) techniques, focusing on improving accuracy, reducing false positives, and enabling real-time detection in dynamic network environments. Several advanced DL techniques are introduced to address these challenges. Entropy-Based Traffic Filtering (ETF) measures the randomness in network traffic to identify anomalies and malicious patterns, reducing noise and improving feature extraction. Self-Supervised Learning for Anomaly Detection (SSLAD) detects malware without labeled data by learning normal traffic patterns and identifying anomalies, thus improving the detection of unknown threats. Graph Neural Networks for Malware Traffic Classification (GNN-MTC) model network traffic as graphs, where devices are nodes, and communications are edges, capturing relational dependencies and anomalies to detect complex attack patterns like botnets and command-and-control (C2) communications. Context-Aware Graph Attention Networks (CA-GAT) further enhance detection by analyzing traffic as graphs while incorporating contextual factors like time and behavior, focusing on relevant interactions to improve attack detection. The proposed DL model achieves 98% accuracy, surpassing DeepMAL (95%) and an entropy-based method by Huang et al. (97.3%). Its strong precision and recall demonstrate superior performance in detecting known and novel malware, making it well-suited for real-time network security applications. The model was implemented using Python. Future research could focus on integrating real-time adaptive learning models, exploring hybrid DL architectures, and enhancing cross-platform malware detection, ensuring scalability and robustness in evolving network security environments.

With the digitization of medical information, illegal activities such as medical crimes and insurance fraud through tampering have increased. Medical images are particularly vulnerable due to their nature as soft copies and their transmission over networks. National research institutions such as NIST provide guidelines that define security control elements for managing medical images, primarily out of concern for system vulnerabilities. However, there is still a lack of established or standardized digital forensic methodologies specifically tailored to the medical imaging domain. This study proposes a digital forensic technique for detecting manipulation in medical images. Two widely adopted PACS (Picture Archiving and Communication System) platforms were selected, and a dataset comprising 82 samples across 40 types of tampering scenarios was constructed. Tampering behaviors such as the editing or deletion of DICOM files were categorized, and forensic analysis of DICOM tags and system artifacts enabled identification of the type and origin of changes. An automated detection module was developed and tested on 110 validation cases. The results demonstrated accurate detection in all instances, depending on whether the changes were reflected in the actual DICOM files. This research marks the first digital forensic approach to medical image tampering detection and is expected to serve as a foundation for future investigative techniques in response to medical-related crimes.

The ocular-motor deception test (ODT) uses pupil dilations and reading behaviors to detect deception. It questions examinees on two illicit activities but classifies deception for only one. Previous studies demonstrate that the ODT discriminates between truthful and deceptive individuals with over 80% accuracy. The present study considered a four-topic screening test and evaluated whether ocular-motor measures could discriminate between truthful and deceptive individuals and pinpoint the specific topic(s) answered deceptively. We recruited 180 participants from the community and the University of Utah. Sixty participants stole $20 (cash), 60 participants stole $20 and a VISA gift card (cash + card), and 60 participants were innocent (innocent). We asked participants about their involvement in four mock crimes: theft of $20, theft of a VISA gift card, vandalism of a parking kiosk, and filing a false police report. Cash participants were deceptive regarding cash statements, cash + card participants were deceptive regarding cash and card statements, and innocent participants were truthful regarding all statements. The computer compared reactions to cash, card, and vandalism statements to those on false report statements to detect deception. As predicted, cash and cash + card participants showed significant changes in ocular-motor measures to cash and both cash and card statements, respectively. A logistic regression model correctly classified 83.3% of innocent participants, 91.7% of cash participants, and 85.5% of cash + card participants. The model correctly classified 86.1%, 81.1%, and 88.3% of answers to the cash, VISA card, and vandalism items, respectively. The findings suggest that a multiple-issue ODT could be valuable in screening applications.

Research shows that digital forensic examiners experience high stress levels due to the nature of their jobs involving exposure to disturbing media. This study conducted a needs analysis by examining digital and multimedia forensic examiners' psychological well-being, coping mechanisms, social support, and attitudes toward and experiences with barriers to counseling and mental health support. Ninety-four digital and multimedia forensic examiners (DFE) completed the anonymous online survey. Respondents were also asked to self-report their primary duty within digital forensics (e.g., image, audio, or video analysts) and whether they were also working as an investigator/detective; 53 were DFE-only, and 41 had dual roles (DFE + detective). Results examined differences in primary duties (e.g., image vs. non-image analyst) and the number of primary duties (e.g., two vs. three). Of the sample, 36% personally sought counseling due to work-related stress. Image forensic analysts reported more psychological distress and barriers toward help-seeking compared with audio and video analysts. 17% (n = 16) of the sample met the diagnostic criteria for PTSD. There were no significant differences between DFE-only and those working dual roles as detectives on psychological well-being and attitudes toward mental health support. Finally, digital forensic examiners who met the diagnostic criteria for PTSD reported 9 out of 15 mental health stigmas, many of which included fear associated with agency culture (e.g., “affect my promotion”). Findings support the need for accessible, agency-supported, and potentially mandated mental health services for DFE to improve well-being and resiliency.

This study explored the effects of flunitrazepam, a benzodiazepine, on the microbiome diversity of Calliphora vicina Robineau-Desvoidy, 1830 (Diptera: Calliphoridae). By examining the microbial shifts throughout developmental stages, the research contributes valuable data to the field of forensic entomotoxicology. Two colonies of 500 adults were fed minced beef liver, spiked and non-spiked with 25 mg of flunitrazepam, and reared under controlled conditions (24°C, relative humidity of 45%, 12:12 light–dark cycle). Following oviposition, egg clusters were transferred, and the experiment was carried out in triplicate under the same experimental conditions. A total of 54 specimens, including all developmental stages, were collected for microbiome investigation via Illumina MiSeq. Both colonies had a 19-day development cycle from eggs to teneral. However, flunitrazepam-fed specimens were heavier, particularly during the pupa and teneral stages. Microbiome analysis revealed significant differences in diversity and composition between the colonies and across developmental stages. Pseudomonadota (Proteobacteria) dominated the control adults, while Bacteroidota and Bacillota (Firmicutes) were more prevalent in flunitrazepam-fed adults. Additionally, Enterobacterales, Lactobacillales, and Morganellaceae showed notable variations across different stages. This study highlights the significant impact of flunitrazepam on the microbiome dynamics of C. vicina, revealing notable morphological changes related to the specimens' weight toward the end of the development cycle and alterations in microbiome composition. These findings have important implications for forensic entomotoxicology, particularly in the accurate estimation of the minimum postmortem interval (mPMI).