MICCAI 2019
Experience
MICCAI 2019 in Shenzhen brought together researchers working across medical imaging, algorithmic fairness, and deep learning for biomedical image analysis. Conference workshops and presentations focused on emerging directions in hypergraphs, domain adaptation, and generative modeling.
Few imp. things at MICCAI
Hypergraphs and their applications in biomedical image analysis
+ The workshop explored hypergraphs for biomedical image analysis, bridging standard graph neural networks with higher-order relational dependencies.
+ The workshop organizers also discussed few interesting applications of hypergraphs in medical domain, like classification using multi modal data and image registration along with low rank data representation
Domain adaptation and GAN
+ This workshop was joint organized by Kamnistas, Jailmer and Anirban, Everyone discussed few of their amazing work on domain adaptation and GAN
+ Domain Adaptation: Active line of research which tries to address model generalization on completely unseen dataset. In case of biomedical data, we train the model on one specific site but when deployment it needs to be generalized to multiple sites
+ Domain Adaptation concepts are more closer towards transfer learning, confounding seperation and optimal transport problems
+ GAN: Interesting research work was presented by Jailmer, which includes (1) MRI2CT, (2) Low res, High res conversion, (3) Progressive approach of training GANs
+ Discussions covered integrating graph architectures with generative adversarial networks for structured cross-modality synthesis.
Industrial lectures
+ Nice to haveVs Must have
+ Medium term to long term analysis
+ Data quality
+ Having add-ons
+ Convince patients not just doctors
Computational Pathology
+ Stain normalization
+ Neuclick talk
Posters
+ Federated learning
+ Universal representation Learning
+ Self inversion networks
Networking
Posters
Fire Dragon Show
Shenzhen sightseeing
My Presentation
Leave a Comment