A recent study in the International Journal of Optics introduces a new spectral reconstruction technique designed to improve terahertz (THz) spectroscopy for non-invasive detection of hazardous substances in mail.
In response to growing security concerns, especially with the surge in global e-commerce, researchers aimed to enhance detection accuracy and reliability in mail screening systems by improving how hidden materials are identified inside packages.
Image Credit: RTimages/Shutterstock.com
Advancements in Terahertz Technology
Terahertz waves, which occupy the frequency range between 0.1 and 10 THz (between microwave and infrared), are uniquely suited for non-destructive inspection.
Their ability to penetrate common packaging materials—like paper, plastic, and textiles—makes them ideal for detecting explosives, narcotics, and other illicit substances, many of which have distinct "fingerprint" spectra in the THz range.
Importantly, THz radiation is non-ionizing, meaning it does not damage the contents being inspected. However, real-world implementation faces challenges. Air gaps and irregular sample positioning inside envelopes can cause reflections and spectral interference, distorting the signal and reducing detection accuracy.
Optimizing Detection: A Novel Reconstruction Approach
To tackle these issues, the research team proposed an advanced spectral reconstruction method that combines Voigt peak fitting with asymmetric least squares (AsLS) baseline correction.
The Voigt function—an effective hybrid of Gaussian and Lorentzian shapes—improves peak resolution, while AsLS helps clean up noise and correct background distortions without affecting meaningful spectral features.
For data collection, the team used a terahertz time-domain spectroscopy (THz-TDS) system, featuring a dual-port femtosecond laser, photoconductive antenna, and off-axis optical setup to achieve high-resolution spectral measurements.
To simulate realistic screening scenarios, the researchers used nalidixic acid and mitomycin—two antibiotics with known THz absorption features—as stand-ins for hazardous substances. These compounds were blended with high-density polyethylene (HDPE), pressed into tablets, and sealed in Express Mail Service (EMS) envelopes. Twenty samples with varying concentrations were prepared to evaluate detection performance.
Principal component analysis (PCA) was used to compare spectral data before and after applying the reconstruction method. The contribution of the first principal component (PC1) jumped from 75 % to over 99 %, a clear indicator of improved signal quality.
Classification models, including support vector machines (SVM) and one-dimensional convolutional neural networks (1D-CNN), were then used to evaluate identification accuracy based on the refined spectral data.
Key Findings: Cleaner Spectra, Better Detection
The outcomes demonstrated that conventional THz spectral measurements were significantly affected by envelope-induced reflections and irregular sample placement, leading to interference peaks and reduced identification accuracy. The reconstructed spectra exhibited enhanced clarity by applying the proposed spectral reconstruction method, effectively eliminating false absorption features and minimizing spectral noise.
Quantitative analysis confirmed substantial improvement in precision, with root mean square error (RMSE) values reduced to 0.31 % for nalidixic acid and 0.28 % for mitomycin.
Classification performance improved notably, with the PCA-SVM model achieving an accuracy of 90.74 %, marking a 7 % increase, while the one-dimensional CNN (1D-CNN) model reached 98.45 %, reflecting nearly a 10 % gain.
These results validate the effectiveness of the spectral reconstruction approach in filtering out packaging-related interferences and enhancing the reliability of THz spectroscopy for mail security screening.
Broader Implications for Security and Detection
This work presents a practical step forward for improving mail screening at postal hubs, customs points, and border security checkpoints. By offering a non-invasive, highly accurate method for detecting hazardous substances inside packages, it reduces reliance on manual inspections and helps streamline security protocols.
Beyond mail security, THz spectroscopy has potential applications in forensic analysis, pharmaceutical authentication, counterfeit prevention, and biomedical diagnostics due to its ability to detect distinct molecular signatures.
The combination of THz technology with advanced machine learning models enhances threat detection while minimizing false positives—an essential improvement for high-throughput, real-time security systems.
Download your PDF copy now!
The Future of Mail Security
This study highlights how spectral reconstruction—specifically, Voigt fitting and AsLS baseline correction—can significantly enhance THz spectroscopy’s performance in real-world mail screening applications. By reducing interference and improving spectral clarity, researchers boosted classification accuracy and demonstrated a clear path toward more reliable, scalable detection systems.
However, challenges remain, particularly in identifying low-intensity or overlapping spectral features in more complex scenarios. Future research should focus on refining signal processing algorithms, improving signal-to-noise ratios, and developing higher-power THz sources for detecting subtler signatures.
As global mail traffic continues to rise, the need for fast, accurate, and automated screening tools grows more urgent. This research not only advances the capabilities of THz spectroscopy but also supports broader efforts to improve public safety and secure the global mail infrastructure.
Journal Reference
Li, T., et al. Identification of Hazardous Substances in Mail by Terahertz Radiation Based on Voigt and AsLS Fitting Spectral Reconstruction. International Journal of Optics, 5636677 (2025). DOI: 10.1155/ijo/5636677, https://onlinelibrary.wiley.com/doi/10.1155/ijo/5636677
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.