CARS Preview 2018

Matija Ciganovic 21 CARS 2018 Preview While the main purpose of the work was registration , another important part was the segmentation of the bone surface from ultrasound. Matija explains: “ We have devised a novel algorithm based on factor graphs. They're graphical models of data that can be used to, in our case, fill in gaps in extracted surfaces, as well as remove false positives. ” Looking back, he says that one thing he has learnt from this work is that sometimes the easiest ways of solving a problem can be the hardest to figure out: “ It’s a situation where a solution is actually staring at you , but maybe you’re thinking in a different direction, and it turns out that the solution that works best is in fact simpler than all the other alternatives that you had in mind. ” His PhD work now is also doing this kind of registration; however, directly focusing on the spine. Prof. Mazda Farshad , a renowned spine surgeon at the Balgrist Hospital , wants to bring the technology into his operating room. Matija tells us about his next steps: “ I’d like to bring some additional machine learning algorithms into this such as deep learning, which is setting new standards in terms of performance in other areas. My hope is to get this technology into the operating room using novel machine learning techniques. ” Would like to know more? Come to Matija’s presentation on Wednesday!