Researchers from Technical University of Munich and Johns Hopkins University Introduce A Vision-Based Robotic System for 3D Ultrasound Imaging

313
Source: https://arxiv.org/pdf/2107.05998.pdf

Ultrasound imaging techniques have proven to be quite valuable for diagnosing a variety of illnesses, including peripheral artery disease (PAD). One of the most frequent disorders amongst the elderly is peripheral arterial disease (PAD), which involves the blockage or narrowing of peripheral blood vessels, limiting blood circulation to certain portions of the body.

Because of their many advantages, ultrasound imaging technologies are one of the most preferred ways to diagnose PAD. Ultrasound imaging is non-invasive, low-cost, and radiation-free, unlike other imaging modalities like computed tomography angiography and magnetic resonance angiography.

The majority of currently available ultrasonic imaging techniques are designed to acquire two-dimensional images in real-time. While this can be highly beneficial in some circumstances, their inability to capture three-dimensional data diminishes the data’s trustworthiness, making them more sensitive to how different physicians applied a procedure. Researchers from the Technical University of Munich, Zhejiang University, and Johns Hopkins University have created a novel robotic system to acquire high-quality 3D ultrasound pictures. This robotic device, described in a pre-publication document on arXiv, could help physicians and healthcare providers collect more accurate anatomical data utilizing ultrasound technology.

As the items being studied cannot be modified while data is being taken, the great majority of 3D ultrasound imaging technologies developed so far do not allow users to capture the whole artery tree of individual human limbs. The quality of the 3D images produced by these systems degrades dramatically if an object or limb moves during the data collection procedure.

Their robotic system is computer vision-based. It starts by using a depth camera to derive the object’s manually planned sweep trajectory. The object’s specific direction is then estimated using the sweeping trajectory. The robotic system developed by Jiang and his colleagues includes a robotic manipulator, a CPLA12875 linear probe, and an ultrasonic interface, in addition to an RGB-D depth camera. The researchers developed a robot operating system (ROS) system, a foundation for building robotic software to control the robotic manipulator’s movements.

The three primary components of the system’s software are a vision-based sweep trajectory extraction approach, an autonomous robotic ultrasound sweep, a 3D ultrasound compounding method, and a movement-monitoring system. The latter is intended to rectify the 3D compounding and update the trajectory retrieved by the system.
Jiang and his colleagues used a custom-designed and gel-based vascular phantom to test the performance of the device they developed. Their results were highly encouraging since their system was able to collect high-quality, comprehensive 3D images of target blood veins even when the item it was investigating was moving.

The preliminary validation on a gel phantom, the researchers said in their study, shows that the proposed approach may yield a promising 3D geometry, even when the scanned object is relocated. Although the proposed technology was demonstrated using a vascular application, it can also be applied for other applications such as ultrasound bone viewing and visualization of bones.

Paper: https://arxiv.org/pdf/2107.05998.pdf

Source: https://techxplore.com/news/2021-08-vision-based-robotic-3d-ultrasound-imaging.html