Forensic scientists have made significant strides in bloodstain identification, leveraging advanced hyperspectral imaging and machine learning to distinguish between human and animal bloodstains with remarkable accuracy.
In a pioneering study published in Applied Spectroscopy, researchers from the Academy of Criminal Investigation, Yunnan Police College, the Department of Forensic Science, Fujian Police College, and the Faculty of Science, Kunming University of Science and Technology have introduced a novel method for identifying bloodstains. The method, which combines hyperspectral imaging technology with the extreme learning machine (ELM) algorithm, has demonstrated superior performance in distinguishing bloodstains from different species, such as human, chicken, and pig blood.
Blood evidence plays a crucial role in solving criminal cases. However, identifying bloodstains at crime scenes can be challenging, especially when suspects attempt to destroy or obscure this vital evidence. Traditional forensic methods for blood identification rely on chemical and biochemical techniques that are often destructive, expensive, and time-consuming. This new approach aims to overcome these limitations by providing a rapid, non-destructive alternative.
Methodology
The researchers utilized a hyperspectral imaging system provided by Shenzhen Zhongda Ruihe Technology Company, Ltd. The system, which operates in the visible-near-infrared (400 to 1000 nm) wavelength range with a 5 nm resolution, captures detailed spectral data from bloodstains. To simulate realistic crime scene conditions, blood samples were smeared onto cotton fabrics of various colors, including white, yellow, blue, brown, red, and black. Human blood was collected from volunteers, while chicken and pig blood were sourced from slaughterhouses (1).
After the bloodstains dried for a month, the hyperspectral data were collected and analyzed using MATLAB 2021b software. The researchers manually extracted regions of interest (ROI) from the spectral data and calculated average spectra. The data were then processed using the ELM algorithm, which was compared to traditional support vector machine (SVM) and random forest (RF) classification algorithms (1).
The ELM algorithm is a type of machine learning model designed for single-layer feedforward neural networks (SLFNs) (2). Unlike traditional learning algorithms, ELM provides a significant advantage in terms of speed and simplicity. It operates by randomly assigning the input weights and biases of hidden neurons, which remain fixed throughout the training process. The key innovation of ELM lies in its use of a least-squares method to determine the output weights, transforming the problem into a linear optimization task. This approach eliminates the need for iterative tuning of parameters, drastically reducing the computational time and complexity associated with traditional neural network training. ELM is particularly well-suited for handling large datasets and high-dimensional data, such as those produced by hyperspectral imaging. Its ability to quickly process vast amounts of information makes it an ideal candidate for applications requiring real-time analysis (2).
Read More: Applications of Hyperspectral Imaging
Findings
The study revealed that the ELM algorithm outperformed SVM and RF algorithms in terms of precision, sensitivity, specificity, and F1 score (a measure of the model's accuracy) (1). This indicates that the ELM algorithm, when combined with hyperspectral imaging, can accurately and rapidly identify bloodstain species. The researchers highlighted the importance of selecting the appropriate number of neurons for the ELM algorithm to optimize training and testing accuracy.
The results emphasize the potential of hyperspectral imaging and machine learning in forensic applications. This approach not only enhances the accuracy of bloodstain identification but also provides a non-destructive and efficient alternative to traditional methods. The ability to quickly and accurately distinguish between human and animal blood can significantly aid criminal investigations, offering new technical references for bloodstain detection and identification (1).
Conclusion
The study by Zhang Jianqiang, Zhang Xinyu, Lin Caiping, and their colleagues represents a meaningful advancement in forensic science. By integrating hyperspectral imaging with the ELM algorithm, the researchers have developed a method that is both rapid and accurate in identifying bloodstain species. This innovative approach addresses the limitations of traditional forensic techniques, offering a promising tool for law enforcement agencies worldwide (1).
In summary, the combination of hyperspectral imaging and the ELM algorithm provides a powerful method for forensic scientists to analyze bloodstains non-destructively. This technology holds potential for broader application in forensic investigations, paving the way for more efficient and reliable crime-solving techniques. The collaborative efforts of the institutions involved in this research highlight the importance of interdisciplinary approaches in advancing forensic science.
References
Jianqiang, Z.; Xinyu, Z.; Caiping, L.; et al. Identification of Bloodstains by Species Using Extreme Learning Machine and Hyperspectral Imaging Technology. Appl. Spectrosc. 2024, 0(0). DOI: 10.1177/00037028241261727
Huérfano-Maldonado, Y.; Mora, M.; Vilches, K.; Hernández-García, R.; Gutiérrez, R.; Vera, M. A Comprehensive Review of Extreme Learning Machine on Medical Imaging. Neurocomputing 2023, 126618. DOI: 10.1016/j.neucom.2023.126618
OSIRIS-APEX: Unveiling the Secrets of Asteroid Apophis Flyby Post-Earth Encounter
September 23rd 2024NASA’s OSIRIS-REx spacecraft embarks on a new mission—OSIRIS-APEX—to study asteroid Apophis after its anticipated close flyby of Earth on Friday, April 13, 2029. Utilizing advanced spectroscopy and other instruments, the mission will provide unprecedented insights into how planetary encounters reshape small bodies in our solar system.
Best of the Week: The Future of Forensic Analysis, Next-Gen Mineral Identification
September 20th 2024Top articles published this week include a preview of our upcoming “The Future of Forensic Analysis” e-book, a few select offerings from “The Future of Forensic Analysis,” and a news story about next-generation mineral identification.