Delivering light to a tumor buried centimeters inside the body has always required threading a fiber-optic cable through tissue to reach it. A research team at Stanford, led by materials scientist Guosong Hong, has now demonstrated a fundamentally different approach: inject nanoparticles into the bloodstream, aim a focused ultrasound beam at the target site, and let the particles convert that mechanical energy into visible light right where it is needed. The results, published in Nature Materials in April 2026, show that the technique works in living mice and produces enough light to activate neurons, raising the prospect of non-surgical photodynamic therapy, light-triggered gene editing, and other treatments that have been held back by the simple problem of getting photons deep into the body.
How the particles work
The nanoparticles, called mechanoluminescent nanotransducers (MLNTs), are tiny crystals engineered to store energy in defects within their atomic lattice. When a focused ultrasound pulse hits them, the resulting mechanical stress releases that stored energy as photons of visible light. The process is called trap-controlled mechanoluminescence, and its key advantage is spatial precision: the particles circulate everywhere in the bloodstream, but they only glow at the exact spot where the ultrasound beam converges. Because modern ultrasound arrays can steer their focal point electronically, the system effectively creates a scannable light source inside the body without any implanted hardware.
The Nature Materials paper backs up these claims with in vivo electrophysiology recordings from mouse brains. Neurons that had been genetically modified to respond to light fired reliably when the ultrasound-activated particles illuminated them. Immunostaining confirmed the activation patterns. Notably, the team archived its spike-sorting analysis code on Zenodo, an open-data repository, so other researchers can inspect and challenge the computational steps behind those neural recordings. That level of transparency is still uncommon in biomedical research and strengthens confidence in the reported results.
A companion protocol paper, published earlier in Nature Protocols and freely accessible through PubMed Central, lays out step-by-step instructions for synthesizing the particles, deploying them in tissue phantoms and circulatory systems, and running behavioral and immunohistochemical assays. The protocol, called deLight, includes equipment specifications and troubleshooting notes, meaning other labs can attempt replication without reverse-engineering the original team’s setup.
What the researchers envision
Hong has outlined three near-term directions for the platform. The first is photodynamic therapy, in which light activates a sensitizer molecule to generate reactive oxygen species that destroy tumor cells. The second involves engineering particle variants that emit ultraviolet light for sterilization or photochemistry applications. The third pairs the internal light source with photoactivatable gene-editing tools, such as Cre recombinase systems that switch on only when illuminated. Those application tracks were described in a Stanford news release and represent the senior author’s own framing of the technology’s potential, not independently confirmed clinical outcomes.
“This could work virtually anywhere in the body,” Hong said in the release, a statement that captures both the platform’s flexibility and the distance still separating it from proven medical use.
The gap between mouse brain and human clinic
Optogenetics experiments, like those in the Nature Materials study, use neurons that have been genetically engineered to fire in response to very low photon counts. Photodynamic therapy is a different challenge entirely. Killing tumor cells typically demands sustained illumination at specific power densities to generate enough reactive oxygen species. Whether circulating MLNTs can produce that level of light output in a human patient, where tissue depths are greater, blood volumes are larger, and particle dilution is more severe, has not been tested.
Focusing ultrasound through human anatomy also introduces complications the rodent experiments did not face. Bones, gas pockets in the gut, and variations in tissue density can distort the ultrasound focal spot. A clinical system would need real-time imaging and feedback to keep the light-generating zone confined to the target and away from healthy structures. The current work demonstrates focal control under controlled laboratory conditions; maintaining that precision across diverse human anatomies is an unsolved engineering problem.
Then there is the question of safety. The nanoparticles must be biocompatible, must clear from the body without accumulating in the liver or spleen at toxic levels, and must not provoke immune reactions at the doses required for adequate light output. No formal toxicology or pharmacokinetic data appear in the published record so far. Those studies are a prerequisite for any regulatory filing and will likely take years to complete.
Readers familiar with cancer research may also wonder how this compares to sonodynamic therapy, an existing approach that uses ultrasound to activate chemical sensitizers directly, without an intermediate light step. Sonodynamic therapy is already in early clinical trials for certain tumors. The MLNT platform would need to demonstrate clear advantages, whether in spatial precision, therapeutic potency, or versatility, to justify the added complexity of a two-step energy conversion process.
A crowded and fast-moving field
Hong’s group is not alone in pursuing ultrasound-triggered light. Other research teams have explored related but distinct approaches, including piezoelectric-to-chemiluminescent energy conversion for deep-tissue molecular imaging and organic nanoparticles that produce persistent luminescence after ultrasound exposure. These parallel efforts use different particle chemistries and different energy-conversion pathways, but their convergence on the same basic idea, that ultrasound can generate useful light at depth, reinforces confidence that the underlying physics is sound.
Which platform, if any, reaches clinical trials first is an open question. For some tumor sites, interventional radiologists can already thread optical fibers close to the target using catheter-based techniques. For brain disorders, implanted optical probes have entered small human trials. Ultrasound-activated nanoparticles promise a non-surgical alternative, but they will need to show clear advantages in safety, efficacy, or cost before displacing procedures that already work. No head-to-head comparisons have been published.
Where the evidence stands in spring 2026
The MLNT platform has cleared important early hurdles. It has a peer-reviewed publication in a top-tier journal with in vivo data, publicly archived analysis code, and a detailed replication protocol. Those are stronger foundations than many early-stage biomedical technologies can claim. At the same time, every question that matters for patients, including long-term safety, effective dosing, manufacturability, regulatory pathway, and comparative effectiveness against existing options, remains unanswered.
The next milestones to watch for are replication by independent labs, studies in animal disease models (particularly cancer), and early toxicology and biodistribution data. If those results hold up, the idea of lighting up a therapy from the inside with nothing more than an injection and an ultrasound wand could move from a compelling laboratory demonstration toward something a physician might actually prescribe. For now, the physics works. The medicine is still catching up.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
