Virtual Ultrasound Images Support Development of Cardiovascular Imaging Tools

Ultrasound imaging can help evaluate the condition, but developing automated diagnostic software requires large numbers of accurately labeled images. Manually outlining vessel walls is time-consuming and can vary between physicians, complicating software training and validation. To provide a more effective alternative, researchers have developed virtual ultrasound images for testing cardiovascular imaging tools.

Daniek van Aarle conducted the research within e/MTIC, the Eindhoven MedTech Innovation Center, a collaboration between Catharina Hospital, Eindhoven University of Technology, Philips, Máxima Medical Center, and Kempenhaeghe. The work focused on computer-generated ultrasound images that closely resemble patient scans. Because the digital blood vessel is built from scratch, researchers know the exact vessel-wall location, vessel size, and degree of movement before software is tested.

The approach is intended to help researchers develop new medical devices, diagnostic software, and artificial intelligence applications. Unlike patient images, virtual ultrasound scans do not require experts to manually define the vessel boundary before use. This allows imaging algorithms to be tested against a known reference standard.

A major part of the work involved patients with abdominal aortic aneurysm. During nearly 30 procedures at Catharina Hospital, physicians captured ultrasound images from inside the blood vessel using a miniature ultrasound probe attached to a thin catheter. Van Aarle then compared these clinical images with virtual ultrasound images generated from medical scans.

The research also connected with earlier work on extracting more information from the vessel wall in patients with abdominal aortic aneurysm. Vessel wall thickness, calcifications, and blood clots may help improve future assessments of aneurysm risk. In that work, simulated ultrasound images complemented data from laboratory studies and patients.

The virtual ultrasound models are still being refined. Van Aarle worked to include subtle biological tissue properties, while highly detailed simulations continued to require substantial computing power. The simulation toolbox developed during the Ph.D. project is now being used in follow-up projects within the research group.

“We can create datasets that would not otherwise exist. That allows us to develop new ways of generating images and extracting information from ultrasound scans,” said Van Aarle.

Related Links
Eindhoven University of Technology