A fiber-optic terahertz camera images living tissue at near video speed

University of Warwick

Most medical imaging forces a trade-off. X-rays penetrate tissue but deliver ionizing radiation. Ultrasound is safe but struggles with surface detail. MRI excels at soft-tissue contrast but demands a room-sized magnet and a patient who can hold still for minutes. Terahertz imaging - using electromagnetic waves that sit between microwaves and infrared light on the spectrum - has long promised a way around these compromises. The waves are non-ionizing, penetrate surface tissue, and are exquisitely sensitive to water content, which differs sharply between healthy and diseased cells.

The problem has always been practical. Terahertz systems tend to be slow, bulky, and tethered to optical benches. Getting one near an actual patient, let alone moving it across a wound in real time, has remained out of reach.

That barrier may be falling. A team at the University of Warwick, led by physicist Emma MacPherson, has built a fully fiber-coupled terahertz system that delivers near video-rate imaging at a spatial resolution of roughly 360 micrometers - more than five times faster than current state-of-the-art platforms. The work, published in Nature Communications, represents the first terahertz imager compact and flexible enough to function as a handheld clinical device.

From optical bench to bedside probe

Traditional terahertz imaging rigs rely on free-space optics: mirrors, lenses, and beam splitters arranged on vibration-isolated tables. Every component must be precisely aligned, and the entire assembly is immovable. MacPherson's team replaced this architecture with optical fibers that carry terahertz pulses from a compact source to a scanning head and back again. The result is a system that can bend, flex, and reach a patient's body.

The approach uses single-pixel imaging, where one detector element captures data while a digital pattern generator controls which part of the scene is illuminated at any moment. Computational algorithms then reconstruct the full image. This is inherently simpler and cheaper than building arrays of terahertz detectors, which remain expensive and difficult to fabricate.

By coupling the entire optical path through fiber, the team eliminated the alignment headaches that have kept terahertz systems trapped in laboratories. The scanning head could, in principle, be mounted on a robotic arm or held by a clinician like a dermatoscope.

Distinguishing fat from protein in living tissue

In proof-of-concept tests, the system successfully differentiated fat and protein regions in porcine tissue samples. The contrast arises because different biological tissues absorb and reflect terahertz waves differently depending on their water content and molecular composition. Fat, with its lower water content, produces a distinct terahertz signature compared to muscle or connective tissue.

More strikingly, the team demonstrated live imaging on a human volunteer, capturing real-time terahertz images of a wound on the subject's arm. The images revealed tissue boundaries and hydration gradients that would be invisible to the naked eye but carry clinical information about healing status.

The spatial resolution of 360 micrometers is fine enough to detect features smaller than a grain of salt. While that does not approach the cellular-level detail of histology, it sits in a useful range for assessing wound margins, mapping tissue hydration, and potentially identifying the boundaries of skin lesions.

Speed that clinical workflows actually demand

Perhaps the most significant advance is speed. Previous terahertz imaging systems required seconds to minutes per frame, making them impractical for anything involving a moving patient or a clinician who cannot afford to wait. MacPherson's system approaches video-rate acquisition, fast enough that a doctor could sweep the device across a wound and watch the image build in near real time.

The five-fold speed improvement over existing systems comes from a combination of the fiber-coupled architecture, optimized pattern sequences for single-pixel reconstruction, and faster data acquisition electronics. Together, these changes move terahertz imaging from a laboratory curiosity toward something that could fit into the rhythm of a busy clinic.

MacPherson envisions applications in wound assessment, where clinicians currently rely on visual inspection and subjective judgment, and in surgical guidance for skin cancer removal, where knowing the exact boundary of a tumor can reduce the need for repeat operations.

What terahertz cannot yet do

The technology has clear limitations. Terahertz waves penetrate only a few hundred micrometers into wet tissue, restricting imaging to surface and near-surface structures. Deep tumors, internal organs, and anything beneath a thick layer of skin remain invisible. The 360-micrometer resolution, while useful, cannot replace biopsy for definitive diagnosis at the cellular level.

The current study is a proof of concept with a small number of demonstrations rather than a clinical trial. No patient outcomes were measured, and the system has not yet been tested in the demanding conditions of a real surgical theater or emergency department. Regulatory approval, clinical validation studies, and manufacturing scale-up all lie ahead.

Single-pixel imaging also imposes a computational overhead. Image quality depends on reconstruction algorithms, and artifacts can appear if the subject moves during acquisition. Whether the system performs reliably across diverse skin types, wound geometries, and clinical environments remains to be established.

A niche that radiation-free imaging could fill

The clinical appetite for non-ionizing imaging at the tissue surface is real. Dermatologists routinely face decisions about whether to biopsy suspicious lesions, wound care specialists lack objective tools for tracking healing, and surgeons removing skin cancers want to know in real time whether they have cleared the margins. Terahertz imaging will not replace any existing modality wholesale, but it occupies a gap that no current technology fills well.

If the Warwick system can be validated in larger clinical studies and manufactured at a reasonable cost, it could become a complement to dermatoscopy and visual inspection rather than a competitor to CT or MRI. The non-ionizing nature of the radiation means there are no dose limits - a clinician could image the same area repeatedly over days or weeks to track changes.

The next steps, according to MacPherson, include integrating the system with robotic surgical platforms and conducting formal clinical evaluations in wound care and dermatology settings. The jump from a laboratory demonstration to a device that changes patient care is never guaranteed, but the engineering obstacles that kept terahertz imaging on the bench appear, for the first time, to be engineering problems rather than physics problems. That distinction matters.