University of Wisconsin-Madison engineers have developed a new technique for improving the penetration and resolution of ultrasound—an imaging tool that’s not only commonly used in medicine, but also in everything from building inspections to underwater navigation.
“It’s a new imaging framework,” says Chu Ma, an assistant professor of electrical and computer engineering at UW-Madison. “We are able to show an order of magnitude improvement of resolution compared to conventional ultrasonic imaging.”
She and PhD student Jinuan Lin combined a new imaging technique they call spatial mixing with new computational reconstruction algorithms to collect information from far-field wavelengths—allowing ultrasound to peer deeper into an object with better resolution than conventional methods. The two described the technique, which could expand applications for ultrasound, in the Jan. 31, 2025, issue of the journal Science Advances.
In ultrasound imaging, a probe called a transducer transmits high-frequency sound waves into a structure—like the human body, or a building. When the waves encounter an object, they reflect back to the transducer, which then builds an image from the information.
It’s a straightforward, tried-and-true technology—but there is a big tradeoff: The deeper acoustic waves travel into an object, the fuzzier the resulting image becomes. That drop in resolution occurs because of the “diffraction limit”—the theoretical limit on how much detail an ultrasound image can show.
In some cases, however, it’s possible to get around these limits by using specially designed materials, called “controlled labels,” to enhance images. In medical imaging, for example, those labels are usually bubble-filled contrast agents injected into or near a target organ. In building inspection, they can be specially coated metal pellets. These materials bounce the weakest, deepest “far-field” ultrasound waves back to the transducer as much stronger signals—enabling the ultrasound to gather subwavelength signals—those smaller than its actual wavelength. The challenge is that these controlled labels must either remain motionless or be tracked precisely during imaging—and that requirement limits the applications for which subwavelength ultrasound can be used.
In their new technique, which is compatible with current ultrasound hardware, Ma and Lin have developed what they call a “spatial mixing” technique using blind labels. In their system, the contrast material is scattered randomly near the target object—eliminating the need to track the material or keep it in a static location. Instead, the ultrasound’s acoustic waves bounce off the blind-label particles wherever they may be, and an algorithm uses that data to piece together the scattered signals and create an image.
The resulting system, which Ma is licensing and patenting through the Wisconsin Alumni Research Foundation, can produce clearer subwavelength images in many more situations. For instance, as blind label particles travel through the bloodstream, they can be used to image structures like blood vessels. The technique can also be used to identify structures hidden in underwater vegetation or to identify schools of fish in the ocean.
While this breakthrough opens ultrasound to many new applications, it is just the first step on Ma’s research pathway. She is already looking for better materials to use for blind labeling in various applications. At the same time, improving algorithms is key in adding flexibility: With the right algorithms, Ma believes, blind labels will not even be necessary to create subwavelength ultrasound images in some applications. She’s developing methods that can, in place of blind labels, leverage randomly distributed static microstructures already found in the human body, building materials or ocean to reflect ultrasound waves.
“For example, the microstructures of tissue or concrete are existing randomly scattered media that could be used instead of these blind labels,” she says. “Then our next step is to explore other computational algorithms, like deep learning and other AI techniques, to further enhance our capability to extract the encoded information and do image reconstruction.”
While she has not yet licensed or commercialized the system, Ma says she thinks it could be implemented in current ultrasound machines as a software update with no hardware modifications, though some applications might need new contrast agents.
Chu Ma is the Dugald C. Jackson Assistant Professor.
The authors acknowledge funding from National Science Foundation grant CCSS-2237619 and the Wisconsin Alumni Research Foundation.
Featured image caption: PhD student Jinuan Lin (left) and ECE Assistant Professor Chu Ma have developed a new technique that allows ultrasound devices to gather subwavelength images without the need to track specialized contrast materials. Credit: Joel Hallberg.