Terahertz sensor boosts biological imaging speed and contrast

Researchers developed a terahertz focal-plane imaging platform that uses a slot-array metamaterial sensor to improve biological tissue detection. The system increased image contrast about threefold and cut acquisition time from about one day to under 30 minutes, while neural-network processing pushed tissue classification accuracy to 98%.

Why it matters: - Terahertz biological imaging has struggled to balance signal enhancement with speed. - The new platform pairs metamaterial-based field amplification with parallel imaging, making high-contrast tissue analysis more practical. - Faster acquisition could matter for large-area samples and higher-throughput biological testing.

What happened: - Researchers published a study in Opto-Electronic Technology on a terahertz focal-plane imaging system based on a slot-array metamaterial sensor. - The sensor was designed and fabricated to resonate at 1.0 THz. - The approach combines metamaterial enhancement with focal-plane imaging and deep learning-based image processing.

The details: - Numerical simulations showed pronounced localized electric-field enhancement at resonance, with sensitivity improved by up to 3.9 times. - Biological samples were placed in direct contact with the metamaterial surface so the enhanced near-field region could interact with tissue. - Compared with a bare silicon substrate, the metamaterial sensor produced about a threefold gain in image contrast. - Plant tissue tests resolved primary veins, secondary veins, and mesophyll regions more clearly. - Bare silicon images could not distinguish fine secondary veins as well. - Cross-sectional profiles and statistical analysis confirmed the contrast gain was nearly threefold. - Animal tissue tests on mutton, pork, and beef showed clearer differences between adipose and muscle tissues. - Data collection time fell from about one day in conventional raster-scanning systems to less than 30 minutes. - A multilayer perceptron model classified pixels using 128-dimensional equivalent attenuation parameter spectra. - The model reached 98% classification accuracy for both plant and animal tissue datasets. - Threshold segmentation and connected-component post-processing made tissue boundaries more distinct and reduced background noise. - The contrast-to-noise ratio improved compared with raw amplitude images. - The work was supported by the National Natural Science Foundation of China, the Beijing Municipal Natural Science Foundation, and the Sino-German Mobility Program of the Sino-German Center for Science Funding.

Between the lines: - The study addresses a core limitation in terahertz biosensing: enhanced sensitivity often comes with slow, point-by-point scanning. - Focal-plane acquisition shifts the field from lab-style proof of concept toward a workflow that could be more usable in real testing settings. - The neural network adds a second layer of improvement by extracting information from the terahertz spectrum rather than relying only on image brightness.

What's next: - The authors position the method as a path toward more efficient, high-sensitivity terahertz biological diagnostics. - The team’s ongoing focus on THz imaging and metamaterial devices suggests further work on sensor design and computational analysis. - The paper is identified as Zhang C, Wang XK, He ZH et al., Biological testing with terahertz focal-plane imaging based on a slot metamaterial sensor, Opto-Electron Technol 2, 260012 (2026), DOI: 10.29026/oet.2026.260012. - More information is available in the journal announcement.