The proposed project aims at developing an autonomous robotic system for performing three different clinical case studies for vascular ultrasound examinations, i.e. a) detection of abdominal aortic aneurysm (AAA), b) estimation of central venous pressure (CVP) from the ultrasound assessment of the jugular venous pulse and c) the compression ultrasound (CUS) maneuver for detecting deep vein thrombosis (DVT).

1. Firstly, we aim at using a collaborative 7-Dof manipulator (KUKA LBR-Med) for developing a system that can autonomously perform the ultrasound screening of the abdominal aorta, measuring its diameter to provide an indicator of aortic aneurysms (AAA). The proposed robotic system could replace the expert physician in performing ultrasound mass screening by moving the probe over the patient and recognizing and measuring the aorta in the ultrasound image. It employs a collaborative robot for probe movement and an ultrasound device for imaging. The image is processed in real-time by a neural network trained on thousands of manually segmented images to identify the aorta. Once the aorta is detected, its coordinates are calculated and used in a movement planning algorithm to optimize its visualization for the diameter measurement. 

2. Then, we will exploit the sensing capabilities of the robot to better understand the interactions between the ultrasound probe and the human tissues. Indeed the second challenge that we will address is the generation of an indirect measure of the central venous pressure (CVP) from the ultrasound assessment of the jugular venous pulse. The acquisition of such a clinical parameter currently requires an invasive procedure (endovenous catheterization) which poses risks to the patient. Recent studies have demonstrated that the same parameter could be estimated by monitoring the variation of the cross-sectional area of the jugular veins, that can be visualized through ultrasound imaging, taking care of applying the proper forces to the neck tissue, so to not deform the vein and invalidate the measure. 

3. Finally, the third clinical challenge that will be addressed is the compression ultrasound (CUS) maneuver: a diagnostic technique for detecting deep vein thrombosis (DVT). It involves an ultrasound probe to apply pressure on the veins, typically in the legs, to observe their compressibility. In healthy veins, the vein walls collapse under pressure, while in veins with thrombus (blood clots), they do not. This non-invasive method is widely used due to its effectiveness and safety, but it is typically performed by expert physicians, because it requires a deep knowledge of the vascular anatomy, the correct application of forces and an accurate image interpretation. Therefore, we propose to develop an autonomous robotic system to perform the CUS procedure. The robot equipped with an ultrasound probe will accurately position and apply the necessary pressure on the patient's leg, guided by real-time AI ultrasound image processing and force feedback. Our algorithms will allow the robot to analyze ultrasound images during the controlled compression and detect signs of deep veins thrombosis.