The course will be divided into lectures during the morning and hands on assignments in the afternoon. Parts of the course will be split into clinical and technical tracts, depending on your level of expertise. Participants of the hackathon are encouraged to come with their data and we will organize (if possible) matching data for validation from other participants on the course.
Our starting point is an overview of the history of Medical Imaging Artificial Intelligence we then discuss the success stories but also the pitfalls. Next, we will review the process from data acquisition, access to the DICOM objects, feature extraction, machine learning (including new developments with Deep Learning) analysis and validation.
The clinical tract will then learn more about the clinical implementation of quantitative imaging, from acquisition protocols to software solutions and finally the implementation of decision support systems.
The technical tract will focus on advances in synthetic data generation and harmonization techniques, new Deep Learning architectures, and current workflow solutions.
In the final part of the course, we will discuss the current challenges and directions of research in the field; in particular, the necessity of dealing with large annotated data sets, the FAIR principles and the distributed learning approach.
Participants are encouraged to bring their datasets for analysis during the hackathon. The dataset has to be fully open source (e.g. from TCIA) or anonymised and cleared by ethics (a written prove of this will be required). If requested in advance, the organisers will perform “data matching” for attendees to facilitate external cross validation. Please contact us to check the availability of this service.
If you want to share your data with Maastricht University during the course you can fill out and sign the DTA template provided here (dta dec18-BD4I Course TEMPLATE). You can also follow the course by storing the data you bring, on your own device, in this case a DTA is not necessary.