One important drawback of today’s clinical care is the transport of patients, often over long distances to receive clinical care at specialized units. Medical imaging is a cornerstone of modern medical diagnostics but is often limited in remote areas. Today, newly developed handheld ultrasound (US) devices exist that can be put in use at remote areas. But a lack of familiarity with US and its non-intuitive nature limits today uses. We propose to increase US diagnostics in remote areas by using artificial intelligence (AI) to guide it.
Guided US will allow new users to confidently use US where the patient lives. As a test case we will develop solutions for detecting abdominal aortic aneurism (AAA). AAA is a gradual dilation of the abdominal aorta, and if left untreated, it may rupture with a high risk of fatal consequences. With AI based US guidance, general practitioners in remote areas can supplement regular AAA management at centralized units, including acute detection of symptomatic aneurysms, screening for asymptomatic AAAs in defined risk-groups, and monitoring growth until treatment.
The project will:
- Create knowledge on users' needs to point-of-care US in remote areas primary care.
- Create datasets for machine learning purposes.
- Develop intelligent US image-interpretation guidance by machine learning based segmentation of key anatomical structures.
- Develop user-interface and real-time functionality.
- Test the prototypes at the three participating municipalities to collect real-life experiences creating an understanding of actual changes in practices as they uncover in clinical use.
