"Reproducibility and Federated Learning"

Deep learning methods are becoming increasingly popular in medical image analysis. Their performance, however, heavily depends on availability, diversity, and quality of the data.

In this episode, Sanam Assili and Egor Panfilov discuss with Francesco Santini current issues in medical image segmentation, federated learning as a promising solution, and recently introduced open source Dafne framework.

Dafne is a joint effort by the University of Basel, the Basel University Hospital, and the Mondino Foundation, more information is available here: https://dafne.network/

Dafne Tutorial - Introduction: https://www.youtube.com/watch?v=TitUKqnXXJw

Outline:

00:00 Introduction

01:23 Segmentation and reproducibility

05:58 Federated learning introduction

07:08 Legal / Data

15:40 Commercial

18:30 Dafne framework

28:40 About MRI together workshop