Computer Vision News - January 2020

3 Summary Smart Ultrasound I aging 1 the acquisition itself, the robustness of ultrasound as an imaging modality is improving. This will no doubt lead to better patient outcomes, but Kristin says the biggest breakthrough she has seen in this area recently is the use of reinforcement learning in terms of the algorithms themselves: “Reinforcement learning has been a big step forward. This is where we have the potential to develop algorithms that can outperform humans. The algorithms that have been used in previous years were designed in such a way that they could reach human performance, but due to being translated into clinically useful tools . “I think the funding bodies are pushing for this kind of collaboration. At GE, we’re initiating a collaboration between universities and hospitals where there is a requirement to spend half of the time in industry . That’s a step forward, but it also requires a change in mentality from all sides to think beyond our own priorities to the common goals and how we can reach those together. It requires a bit of compromise on each side, but that’s what is needed in order to really make this happen.” the way they were structured, could not outperform it. With this being more and more widely used, the accuracy and the usability of these types of algorithms will increase quite substantially over time.” Thinking ahead to the next 12 months, Kristin would like to see more collaboration between the academic and industrial sites , andmore clinicallydrivenalgorithm development. In the past, universities have developed advanced algorithms to solve problems that are not necessarily suitable within the clinical workflow, so she’d like to see that academic work Great image quality on GE Healthcare's 4Vc ultrasound probe