Sonar and ultrasound, which use sound as a navigational device and to paint accurate pictures of an environment, are the foundation of many technologies, including medical ultrasound units and military navigation systems. However, when it comes to more effective sonar and ultrasound, animals’ biosonar capabilities still are far more superior.

In a new study that examined bats, dolphins, and mole rats, Prof. Nathan Intrator of Tel Aviv University’s Blavatnik School of Computer Science (Israel), in collaboration with Brown University’s (Providence, RI, USA) Prof. Jim Simmons, is working to identify the reasons biosonar has the advantage over man-made technologies. Using a unique method for measuring how the animals interpret the returning signals, Prof. Intrator has determined that the key to these animals’ success is better, real-time data processing. “Animal ‘echolocations’ are done in fractions of milliseconds, at a resolution so high that a dolphin can see a tennis ball from approximately 80 meters away,” he said, noting that the animals are able to process several pieces of information simultaneously.

Their research, which was reported in the Journal of the Acoustical Society of America and presented at the 2010 and 2011 Machine Learning for Signal Processing (MLSP) conferences, could lead to cutting-edge navigation systems and more accurate medical imaging.

Biosonar animals send ultrasonic sounds called “pings” into the environment. The shape of the returning signals, or echoes, determines how these animals “see” their surroundings, helping them to navigate or hunt for prey. In a matter of tens of milliseconds, the neurons in the animal’s brain are capable of a full-scale analysis of their surroundings represented in three dimensions, with little energy consumption. Even with the help of a supercomputer, which consumes thousands of times more energy, humans cannot generate such an accurate picture, Prof. Intrator reported. Using echolocation, a bat can tell the difference between a fly in motion or at rest, or determine which of two fruits is heavier by observing their movements in the wind.

Fascinated by the quality of the natural biosonar over its man-made counterparts, Profs. Intrator and Simmons set out to examine how biosonar animals perform echolocation so rapidly and accurately. Using an electronic system, they modified the frequency and noise levels of the echo returned to the animal. By manipulating the echo, the researchers could determine what factors of the returning signal reduced an animal’s ability to accurately analyze the returns. This then led to a better determination of how the returning echoes are represented and analyzed in the animal’s brain.

Prof. Intrator and his fellow researchers have created mathematical models, involving machine signal and learning processing, which improve man’s ability to interpret the echoes. This will lead to more accurate echo localization and better resilience to background noise. Once researchers collect more data about animal interpretation of biosonar, they will be able to mimic this technology for better ultrasound and sonar systems, according to Prof. Intrator. “Animals explore pings with multiple filters or receptive fields, and we have demonstrated that exploring each ping in multiple ways can lead to higher accuracy,” he explained. “By understanding sonar animals, we can create a new family of ultrasound systems that will be able to explore our bodies with more accurate medical imaging.”

This new development could provide a variety of benefits to the medical imaging field, such as earlier detection of defects in embryos or noninvasive detection of cancer tumors. Dissimilar to a magnetic resonance imaging (MRI) or computed tomography (CT) scanner, which are large, costly to operate, and frequently use hazardous radiation, the new generation of ultrasound machines could be used in a physician’s office at a portion of the cost.

Related Links:
Blavatnik School of Computer Science
Brown University