MICCAI 2018 Daily - Wednesday

ThinkSono are a start-up based in London and Berlin. They are an interdisciplinary team, from a range of different nationalities and professional backgrounds, working on the problem of deep vein thrombosis diagnosis. Sven Mishchkewitz, the co- founder and CTO, and Antonios Makropoulos, the Chief Scientific Officer, speak to us about their poster yesterday - AutoDVT: Joint Real-time Classification for Vein Compressibility Analysis in Deep Vein Thrombosis Ultrasound Diagnostics Their work is a system to automatically diagnose deep vein thrombosis (DVT). DVT is a blood clot in your leg and if it breaks loose, you can get a pulmonary embolism, which is very deadly. They are attempting to provide point-of- care diagnostics to solve this problem. The diagnosis is done via ultrasound and they are presenting real-time ultrasound analytics. Antonios explains there are two main parallel tasks to solve. One is to find the position that you need to scan – the landmark. The second is to detect whether the vein closes or not, which means whether you have a clot or not. But why is such an indispensable tool not available already? Sven says the answer is timing. The tools, the algorithms, the knowledge, and also, the hardware that you need to do this at the point of care has only become available in the last couple of years. Research in the field and products attempting to use ultrasound are just coming out. He points out that DVT is a very niche topic as well. A lot of people are looking at cardiac and other diseases, but thrombosis is not such a sexy topic. Antonios tells us that ultrasound has become more popular recently. It is inherently a much noisier acquisition than MRI or CT, so up to now, people were optimising their models with better images. However, with deep learning tools, you can now more easily do research with noisier images and get better predictions. 12 Poster Presentation With Sven Mishchkewitz and Antonios Makropoulos “ With deep learning tools, you can now more easily do research with noisier images and get better predictions! ” Wednesday