Kosmiko: From CMOS Imaging Sensors to Pixel-Calibrated Particle Instruments

The sky above us is never truly empty. Even on a clear, sunny day, a constant rain of invisible particles from deep space and from the decay of natural elements within the Earth itself is passing through everything. These are cosmic rays and other forms of ionizing radiation. When they strike matter, they leave behind a faint trace of energy, a tiny spark that can be detected if one knows where to look. For decades, scientists have used expensive, specialized equipment to measure this invisible rain, but such instruments are heavy, costly, and difficult to deploy in large numbers. The question has long been whether the cameras we already carry in our pockets or mount on our roofs could be coaxed into seeing this same radiation, turning everyday devices into a vast, distributed network of sensors.

A team of researchers in France has answered that question with a new, low-cost system they call Kosmiko. Instead of building a complex machine from scratch, they built a detector using a small, affordable computer board and two standard camera modules, the kind often used for high-quality video. They placed these cameras in complete darkness, removed their lenses and filters, and programmed them to stare into the void for hours at a time. The goal was not to take pictures of the world, but to listen for the tiny electrical sparks that occur when a single particle of radiation hits the camera's silicon sensor. By treating each tiny dot on the sensor as its own independent detector, the team created a method to spot these events without being overwhelmed by the camera's own internal noise.

The process begins with a careful calibration. Before the cameras start their long watch, they take thousands of pictures in the dark to learn what "normal" looks like for every single pixel. This allows the system to understand the unique background hum of each dot, distinguishing between a random electronic glitch and a genuine hit from a particle. Once the cameras are ready, they begin a twelve-hour run, capturing images at a rate of approximately one per second. The software running on the small computer board checks every pixel in every frame against its own personal threshold. If a pixel suddenly glows brighter than its usual dark state, the system flags it as a candidate event. Crucially, the system does not save the entire image, which would fill up the memory card in minutes. Instead, it saves only the coordinates and brightness of the pixels that fired, along with the time they fired. This massive reduction in data allows the device to run for days or weeks, collecting a stream of events rather than a mountain of empty pictures.

The researchers tested this method in various locations to see what the cameras would find. They started at sea level, where the rate of detected events was high, as expected. They then moved the device deep underground, to a laboratory shielded by 1500 meters of rock and water. They anticipated that the rate of events would drop dramatically, as the rock would block most of the cosmic rays coming from space. Surprisingly, the rate did not fall as much as they hoped; the cameras continued to register a steady stream of hits. This led the team to investigate what else might be causing the signals. They realized that the camera itself, specifically the glass lens and the infrared filter sitting right in front of the sensor, contained trace amounts of natural radioactive elements. These elements were decaying and firing particles directly into the sensor, creating a background noise that masked the true cosmic signal.

When the researchers removed the lens and the filter, leaving the sensor exposed, the results changed instantly. In their underground tests, the event rate dropped significantly, confirming that the camera's own components had been a major source of the false signals. With these parts gone, the device became a much cleaner instrument. To prove that the remaining signals were indeed from radiation, the team exposed the cameras to known sources of gamma rays and alpha particles. The cameras responded with distinct patterns: the gamma rays created small, scattered clusters of bright pixels, while the heavier alpha particles left behind larger, compact blobs. These patterns were recorded as specific morphological classes—such as "track-like" or "compact cluster"—rather than as definitive identifications of specific particle types. This confirmed that the system was working as intended, capable of categorizing ionizing events based on the shape and size of the clusters they left behind.

The final phase of the study involved a large-scale test of reliability. The team deployed twenty of these devices, totaling forty cameras, across six different locations in France, ranging from the coast to high mountain altitudes. They found that the devices produced consistent results wherever they were placed. A camera in one city gave the same reading as an identical camera in another city at the same time, proving that the method is reproducible and robust. While the exact number of events varied from place to place, likely due to differences in altitude and local geology, the system proved capable of mapping the invisible radiation environment with a level of detail previously impossible for such low-cost tools.

This work does not claim to identify every specific type of particle or to provide a perfect measurement of the cosmic ray flux. The sensors are too thin to catch every particle that passes through, and the system cannot yet distinguish between a cosmic ray and a particle from a local radioactive source without careful calibration. However, the study successfully demonstrates a new way to observe the natural radiation environment. By turning a standard camera into a calibrated particle detector, the researchers have shown that it is possible to build a network of sensors that is cheap, easy to deploy, and scientifically useful. The method offers a practical path forward for monitoring radiation levels in diverse environments, from the surface of the Earth to the depths of underground laboratories, using technology that is already available to everyone.