After more than three years of investigation, researchers at the University of Wisconsin-Madison have created an improved approach to acoustic imaging.
In a project headed by electrical and computer engineering professor Chu Ma and Ph.D. student Jinuan Lin, the team combined research into improved hardware with research into improved software to better this imaging at greater distances into a medium, such as the human body.
One of the key pieces of hardware in acoustic imaging, according to Ma, is the transducer, which converts electrical signals into sound waves and then converts them back into an electrical signal that creates the image. Because of this, many labs attempt to increase the capabilities of these transducers. Other labs focus solely on the software producing the images.
“Our uniqueness is that we do it in both,” Ma said. “We took both approaches and combined them to show that it is better than just one of them.”
Ultrasound images lack detail once the waves from the transducers travel too far into the body. The “tradeoff” between resolution and penetration was previously very difficult to overcome.
“If people want to have high resolution, then they want to use high frequency because high frequencies have small wavelengths, and the wavelength usually constrains the resolution of the imaging system,” Ma said. “But high frequency waves cannot propagate long distances in human tissue, in water or in whatever medium.”
These acoustic imaging techniques work because of particles called “blind labels.” As Ma and Lin both described, blind labels are randomly scattered particles smaller than the wavelengths entering the system and placed around the object of interest so that the imaging systems have motion to pick up. Improving the capabilities of these blind labels was a key breakthrough in Ma and Lin’s work.
“The blind labels, because they distribute it around the objects, will provide the spatial mixing to the object,” Lin said. “And that is the key reason they can enable the subwavelength imaging.”
Random particles like these were already used in acoustic imaging. When looking underwater, Ma said, even schools of fish can be used as randomly scattered particles because they can send signals back to the transducer so an image can be produced. What was critical to this research project, in part, was increasing the use of artificial labels.
“Because acoustic imaging is not just for biomedical imaging, it can also be used for non-destructive testing, for example, inspecting pipes,” Ma said. “Then we can flow the particles through the pipes or underwater.”
Going in, it was unclear whether this type of development would even be possible, Lin said. But after testing out various aspects of the work in other mediums like air, the team became more confident their findings would be applicable to several important areas.
“This kind of blind label imaging framework works in air, and we publish that journal paper in ‘Physical Review Applied’,” Lin said. “Then we want to expand application scenarios, so naturally thinking, ‘what if we did it underwater or in the human body?’”
The use of this research outside of the lab is not immediate. Lin said even after the majority of the research was completed it still took about a year to finish revising, which was longer than it typically takes for the prior research she has done. Before this blind label technique can be applied in fields like biomedical imaging, the researchers will have to demonstrate its effectiveness to those who would use it.
“Yes, we have this technology now,” Ma said. “It’s currently a prototype, but we will work together with those clinicians and the people in hospitals and also other areas, specifically in nondestructive testing or underwater sonar.”
