CREATE-MIA Day at the University of Sherbrooke

The September 25 CREATE-MIA Day is being organized by Prof. Max Descoteaux in Sherbrooke, Quebec with talks and presentations both at the University of Sherbrooke Computer Science Department and at Espace Inc., a startup incubator and accelerator.

7:35am - Montreal-based CREATE-MIA participants are to meet at the Roddick Gates of McGill University at the intersection of Sherbrooke O. and McGill College  in preparation for departure by bus at 7:45am.

MORNING SESSIONS: Espace Inc. 400 rue Marquette, Sherbrooke, QC

10h00 - 11h00 Joel Sirois, BioIntelligence Technologies.

11h00 - 11h30 Max Descoteaux

11h30 - 12h Tour of Espace Inc.

12h15 - Departure (by whichever mode of transport you arrived here by) for the University of Sherbrooke.

AFTERNOON SESSIONS: University of Sherbrooke Faculty of Science, Computer Science Department, 2500 boul. de l'Université.

12h30-13h30 LUNCH

13h30-14h30 Prof. Martin Lepage - Small animal vascular imaging in cancer research

14h30-15h30 Prof. Pierre-Marc Jodoin - Medical imaging segmentation using state-of-the-art computer vision and machine learning

15h30 - 16h00 COFFEE BREAK

16h00 - 17h00 Prof. Pierre-Michel Bernier - Neuroimaging methods to study motor control

17h15 Departure by bus back to Montreal with expected arrival at approximately 19h15.

Title: Small animal vascular imaging in cancer research

Speaker: Prof. Martin Lepage, Institute de pharmacologie de Sherbrooke, Université de Sherbrooke

Abstract: The vascular system is part of every human disease and contributes relevant information about the human brain function. Detection of very small blood vessels may be achieved by magnetic resonance imaging but there are still limits and our view of the vascular system is still biased towards larger vessels. Characterization of the blood vessel permeability is even more challenging as a high temporal resolution is required in addition to a high spatial resolution. Strategies involving rapid image acquisition and the combination of microscopic with macroscopic information have been recently developed.

Title: Deep convolution neural nets for segmenting medical images : why are we facing a new revolution?

Speaker: Prof. Pierre-Marc Jodoin, Faculté des Sciences, Informatique, Université de Sherbrooke

Abstract: In the last 2 years, a growing number of medical image segmentation challenges have been won by deep learning methods.  In this presentation, I will revise what deep convolutional neural nets are, how they work, and how they can be applied to segment medical images.  I will also emphasize what makes these methods so successful and why they are more than just the flavor of the month.

Title: Neuroimaging methods to study motor control

Speaker: Prof. Pierre-Michel Bernier, Faculté des sciences de l'activité physique, Université de Sherbrooke

Abstract: Although the cortical regions that participate in human motor control are fairly well characterized, the mechanisms by which these regions interact during movement are still unknown. This talk will present some recent methodological advances that have allowed to better capture the spatio-temporal dynamics of neural activity in motor networks, and as a result provided valuable insights into the functional contribution of these regions.