Researchers improve medical imaging technique

Optoacoustic imaging is still a relatively nascent technique for medical imaging, a press release issued by the Swiss Federal Institute of Technology in Zurich (ETH) states. This technique can, for example, be used to visualize blood vessels inside the human body or to diagnose breast cancer. The image quality is highly dependent on the number of sensors used. Researchers from ETH and the University of Zurich (UZH) have now improved the technique so that the number of sensors can be reduced, all the while maintaining the image quality. With this new method, it should now be possible to increase imaging speed, improve the accuracy of diagnoses and manufacture more cost-effective devices.  

For optoacoustic imaging, “very short laser pulses are sent into the tissue, where they are absorbed and converted into ultrasonic waves”, ETH explains. The ultrasonic waves are used to produce an image in a similar fashion to ultrasound imaging. The team headed up by Daniel Razansky, Professor of Biomedical Imaging at ETH Zurich and the University of Zurich, originally used a self-developed optoacoustic device with 512 sensors in their experiments. The high-quality images this produced were analyzed by an artificial neural network, which was able to learn the features of the high-quality images. The number of sensors was then reduced to 128, then further down to 32. This resulted in “streak-type distortions” appearing in the images. However, Artificial Intelligence (AI) was able to complete the missing data and correct these distortions. In this way, the reduced number of sensors was able to produce an image quality similar to that generated by 512 sensors.

In fundamental terms, the quality of optoacoustic imaging also depends on the focus object being captured from as many different angles as possible. With the newly developed algorithm, it has been possible to improve the quality of images that were recorded from just a narrowly circumscribed sector. “This is particularly important for clinical applications, as the laser pulses cannot penetrate the entire human body, hence the imaged region is normally only accessible from one direction”, Razansky commented. He also indicated that the new method is not limited to merely improving the quality of optoacoustic imaging, explaining that: “You can basically use the same methodology to produce high-quality images from any sort of sparse data”.