.png){kind=logo}
Scientists at ETH Zurich and EPFL have developed an AI-powered camera system that can reconstruct the paths of subatomic particles using only a handful of photons.
The technology, called PLATON, could simplify particle detectors by replacing millions of conventional detector elements with a single light-emitting block and artificial intelligence. The findings were published in Nature Communications.
Particle detectors help scientists study elementary particles by tracking how they move through matter. Most existing detectors use scintillators, materials that emit tiny flashes of light when struck by charged particles.
To pinpoint where these interactions take place, the scintillator is divided into millions of small segments connected to optical fibres and photon sensors. While this design offers high precision, it also increases the complexity and cost of large-scale detectors.
Experiments such as Japan's T2K neutrino project and CERN's LHCb rely on these segmented detector systems to accurately track particle interactions.
PLATON removes the need for millions of detector elements. Instead, it uses a single, unsegmented scintillator block and reconstructs the particle's trajectory by analysing the emitted light.
The system combines a plenoptic, or light-field, camera with a single-photon avalanche diode (SPAD) sensor that detects individual photons generated inside the scintillator. Unlike conventional cameras, a light-field camera records both the intensity of light and the direction from which it arrives.
The research team built the prototype by mounting a micro-lens array directly on the SwissSPAD2 imaging sensor. The setup captures extremely faint light signals while filtering background noise through gated photon detection.
During laboratory tests, the prototype reconstructed particle interactions using as few as five detected photons, with the experimental results closely matching computer simulations.
Also Read: Best Pet Tech Gadgets of 2026 Compared: AI Feeders, GPS Trackers, Cameras & Health Monitors
The researchers also developed a transformer-based neural network to analyse where and when photons are detected. Instead of processing text, the AI identifies patterns in photon data to reconstruct the original particle path in three dimensions.
Simulations suggest a future version of PLATON could achieve sub-millimetre spatial resolution inside a 10×10×10 cm detector. A larger one-cubic-metre detector is expected to deliver resolution of only a few millimetres, comparable to today's advanced scintillator detectors despite using far fewer components.
The researchers also see potential beyond particle physics. They have filed three patents covering the use of PLATON in positron emission tomography (PET) scanners. The technology could improve medical imaging by reconstructing faint light signals more accurately while reducing the complexity of detector systems.
