Simpler PET Scanners Through Monolithic Crystal Imaging

Positron emission tomography sits near the ceiling of diagnostic imaging costs. High-end clinical systems routinely run between 1.5 million and 3 million euros, largely because covering a human torso requires an array of tens of thousands of individual scintillator crystals, each paired to dedicated light sensors. Every boundary between those components introduces signal loss, which clinicians offset either by lengthening scan times or by administering higher activities of radioactive tracers like fluorine-18 fluorodeoxyglucose.

Particle physicists building the PLATON detector system demonstrate a different physical arrangement: a single, uninterrupted volume of scintillator viewed by a plenoptic camera and fast silicon photomultipliers. Instead of trying to isolate an interaction inside an isolated crystal channel, a neural network reconstructs the exact three-dimensional coordinates and arrival timing of gamma photons from the diffuse pattern of light emitted across the entire block. Simulations indicate the approach matches the sub-millimetre spatial resolution of modular arrays while eliminating thousands of assembly parts.

From hardware segmentation to algorithmic reconstruction

When diagnostic hardware leans on predictive algorithms to fill gaps left by simplified mechanics, verification becomes your first question. If you track high-sensitivity biomarkers that require advanced nuclear imaging, ask the radiologist whether a given clinic runs standard time-of-flight arrays or computational reconstruction pipelines. The distinction dictates not only the tracer dose entered into your body, but the physical limits of what that scan can genuinely verify.