You are here: indexgdr2016

Differences

This shows you the differences between the selected revision and the current version of the page.

gdr2016 2016/11/18 12:11 gdr2016 2016/12/13 07:55 current
Line 66: Line 66:
** 10:00 - 10:45 ** | ** Hassan Alhajj, IT2IM team, LaTIM-Inserm U1101, University of Bretagne and Telecom Bretagne ** \\ ** 10:00 - 10:45 ** | ** Hassan Alhajj, IT2IM team, LaTIM-Inserm U1101, University of Bretagne and Telecom Bretagne ** \\
-__Title__: \\ +__Title__: Video analysis for ophthalmic surgery \\ 
-__Resume__: \\+__Resume__: Data recorded and stored during video-monitored surgeries are a relevant source of information for surgeons, especially during their training period. But today, this data is virtually unexploited. Accordingly, different methods have emerged to help the surgeons in different manners: report generation, surgical skill evaluation, construction of educational videos and real-time video monitoring. We focus on the latter application, with the aim to automatically communicate useful information to the surgeon during the surgery. In particular, our goal is to set up a warning/recommendation generation system for videos recorded during cataract surgeries. To distinguish a normal conduct of surgery from an abnormal one, a crucial step is to recognize surgical tasks, phases or gestures in real-time. So, we worked first on recognizing them in the microscope videos. The results obtained are very encouraging but highlighted one main challenge: to improve the interpretation of the videos, one should be able to detect all surgical instruments. However, these instruments have a wide variety of shapes and are only partially visible in the surgical scene. To overcome this issue, we added a second video stream, filming the operating table. In this context, knowing which instruments exit or enter the operating table leads to which tools are likely being used by the surgeon and which tools surely are not. It’s an in progress work and we are currently trying to solve this task using deep learning techniques. \\
** 10:45 - 11:30 ** | ** Pierre Chatelain, Lagadic Team, IRISA, University of Rennes 1 and Technical University of Munich ** \\ ** 10:45 - 11:30 ** | ** Pierre Chatelain, Lagadic Team, IRISA, University of Rennes 1 and Technical University of Munich ** \\
Line 80: Line 80:
** 14:00 - 14:45 ** | ** Alexandre Thouroude, STAN Institute and Surgical Training School of Nancy, University of Lorraine ** \\ ** 14:00 - 14:45 ** | ** Alexandre Thouroude, STAN Institute and Surgical Training School of Nancy, University of Lorraine ** \\
-__Title__: \\ +__Title__: Learning the optimal gesture, from a Mirage 2000 pilot to surgeon \\ 
-__Resume__: \\+__Resume__: Learning times must be shorter and shorter. Learning costs are more expensive. Robotic surgery is embedded in these immutable parameters. How could we translate the training of fighter pilots to surgeons? The main point is not necessary in the materials but how we use them. \\
** 14:45 - 15:30 ** | ** Russell H. Taylor, Department of Computer Science, The Johns Hopkins University ** \\ ** 14:45 - 15:30 ** | ** Russell H. Taylor, Department of Computer Science, The Johns Hopkins University ** \\
inserm rennes1 ltsi