Researchers Develop Radiation Detector Based on Tetris Prototype
Inspired by the tetramino shapes from the classic Tetris video game, researchers at the Massachusetts Institute of Technology (MIT) in the United States have developed a simple radiation detector that can safely and efficiently track radioactive sources. The device uses a machine learning algorithm to process data, which allows it to create accurate maps of sources.
In a conventional detector, radiation maps are created by monitoring the intensity distribution over a 10×10 array of detector pixels. The main disadvantage is that radiation can hit the detector from different directions and distances, which makes it difficult to extract useful information about the radiation source. Usually, this is done by applying an absorbing mask to the pixels, which provides certain information about the direction, and performing extensive data processing.
For the MIT researchers, the first step to reducing the complexity of this process was to minimize the redundant information collected by multiple pixels in the array. To do this, they use small lead spacers between pixels to increase contrast and ensure that each detector receives clear information, even if the radiation source is at a great distance.
As a next step, they developed machine learning algorithms to obtain more accurate information about the direction of incoming radiation and the distance from the detector to the source. The inspiration for the last stage of development came from an unlikely source. In the game Tetris, players encounter seven unique tetraminoes that represent all the possible ways to combine four squares to create shapes. By using these shapes to create the detector’s pixel arrays, the researchers predicted that they could achieve the same level of accuracy as detectors with much larger square arrays.
To demonstrate this, the team developed a series of four-pixel radiation detectors in which the pixels are arranged in a tetramino shape. To build radiation maps, these arrays were moved in a circle around the radioactive sources under study. This allowed the detector algorithms to recognize accurate information about the position and direction of the source based on the counts obtained by the four pixels.
The successful field testing at Lawrence Berkeley National Laboratory deserves special attention. Even in the absence of the exact location of the source, the machine learning algorithm was able to effectively localize it in real experimental data.
The innovative approach to detector design and data processing can be useful for radiation detection, and successful field trials emphasize the practical applicability of the approach, contributing to the safety and efficiency of radiation monitoring.
The researchers hope that this type of detector can be implemented for monitoring nuclear reactors, reprocessing radioactive materials, and safe storage of hazardous radioactive waste.
According to Physics World