Big Data, Annotations & Machine Learning for Neuro MR: A No-Brainer
Member-Initiated Symposium

ORGANIZERS: Matthew Budde, Alex Rovira, Hugo Vrenken

Thursday, 21 June 2018

Session Number: MIS-14

Overview
This Symposium was proposed by the White Matter study group.

The advance of Open Science and sharing of research data create excellent opportunities for expanding and improving how MRI is used in studying diseases of the central nervous system (CNS). This symposium focuses on three important aspects: sharing of data and methods; image annotations and general public involvement; and machine learning enabled by big data and data sharing.
Challenges in data sharing will be discussed by Dr. Dyrby. Shared datasets have been used successfully in several diseases, as discussed by Dr. Valsasina. By sharing analysis methods, such shared data are made easy to analyze, as Dr. Stikov will discuss. Shared benchmarks and reference data often require labor-intensive image annotations. Dr. Toro will address how the general public can accelerate scientific discovery by partially replacing expensive experts. Finally, Prof. Ourselin will discuss how machine learning can help analyze Big Data to understand diseases of the CNS.
Data sharing and big data contribute to many studies presented in scientific ISMRM sessions in Neuro and beyond; this complementary symposium addresses this as its central theme. Tuesday’s Plenary Session “Challenging the Assumptions of MRI” will undoubtedly discuss machine learning in image acquisition, analysis and interpretation. Also, the Educational Courses “Deep Learning: Everything You Want to Know” (Saturday) and “Machine Learning for Magnetic Resonance in Medicine” (Thursday) confirm the interest in machine learning.

Target Audience
Scientists and clinicians interested in neurological diseases and in the opportunities of new technologies and approaches to science in this respect.

Educational Objectives
As a result of attending this course, participants should be able to:
-Identify possibilities and challenges of sharing data and methods in neuroimaging, including clinical data, meta-data, non-MR imaging data, and analysis methods;
- Name important insights into disease mechanisms obtained from (large) data sharing initiatives;
- Recognize “crowd-sourced” or “citizen science” initiatives that aim to advance neuroimaging studies of the CNS by engaging the general public; and
- Describe state-of-the-art information on the use of machine learning techniques for studies of the CNS.