IMDEA Materials Institute has developed an acoustic imaging method that overcomes some of the fundamental limitations of technologies such as ultrasound imaging.
The system can reconstruct a complete and accurate image using a single fixed sensor that captures one signal at a single instant, eliminating the need for complex multi-sensor arrays or lengthy scanning times.
Conventional acoustic imaging techniques require either a probe equipped with multiple transducers (a sensor array) that simultaneously capture signals from different locations, or a single sensor that must move or perform repeated measurements over time.
“Traditional approaches involve significant complexity, cost, and equipment size,” explains IMDEA Materials Principal Investigator, Prof. Johan Christensen. “The new method resolves this trade-off by encoding all the spatial information of an object into the frequency spectrum of a single sound wave”.
As outlined in a new publication in Advanced Functional Materials, the system operates by emitting a single broadband ultrasound pulse onto an object. The scattered waves, which contain information about the object’s shape, then pass through a low-cost “disordered medium.”
In the experiments, this material consisted of an agar sheet containing randomly distributed steel beads, which acts as a physical encoder.
Due to its random structure, each frequency component of the sound wave travels along a slightly different path, creating a unique and complex spectral “fingerprint” that is captured by a single sensor.
Although unintelligible to a human observer, this spectral fingerprint contains all the information required to reconstruct the image.
The team used an artificial intelligence neural network trained to act as a decoder. Once trained, the AI can interpret the received signal and reconstruct an image of the original object within tens of microseconds.
In testing, the system successfully reconstructed images of handwritten digits engraved on metal plates with high fidelity, achieving structural similarity scores of up to 98.7% compared with the originals.
“Instead of requiring an expensive sensor array, this material encodes all spatial information into the frequency domain, which an AI can read instantaneously. This dramatically reduces the complexity and cost of imaging hardware,” says Prof. Christensen.
This technology opens the door to the development of ultra-compact, low-cost, real-time acoustic imaging devices.
Its potential applications are wide-ranging, from medical diagnostics, enabling the visualisation of moving organs such as a beating heart using much simpler equipment, to non-destructive material inspection in industry and underwater object detection.
The research team notes that the mechanism is universal and could be extended to three-dimensional imaging and to other types of waves.
This work is the result of an international collaboration between the University of Nanjing, Nanjing Forestry University and Wuhan Polytechnic University in China, and IMDEA Materials Institute.