This week’s episode of Reasons to do AI with Friends elicited such reactions from viewers in the live chat as “Thank you so much! It was fascinating!” and “Really impressive work! Thanks for sharing.”
In the fifth episode of ESR Connect Weekly’s Season 2 on artificial intelligence, Dr. Mariana Diogo from the Medical University of Vienna presented some general background on the corpus callosum and corpus callosum agenesis. She then provided an overview of the fetal MRI approach and the effectiveness of T2, DTI, BOLD and spectroscopy in assessing this region. Subsequently, she explained the clinical difficulties in images of the fetal brain and prediction of the development.
Prof. Georg Langs, who has been a great addition in two episodes this season and who is also from the Medical University of Vienna, then explained how AI can help with this particular medical issue. Machine learning can not only help in diagnosis but also in better understanding this very complex disease and how this is a Spatio-temporal modelling task. He then went on to explain the requirements for creating an algorithm that tackles this issue, the types, formats and annotation of the data, in order to teach the model how to segment an infant’s brain. As in previous episodes, Langs also covered pitfalls and challenges in fetal or utero images and how his team tackled them. The audience was then presented details on the data validation in the lab and, last but not least, the clinical validation.
This week’s Theo’s Technical Top Tips segment focused on data annotation. Theo Barefoot, an MRI physicist at the Royal Marsden London, presents one valuable detail to the viewers in each episode. Barefoot explained machine learning terms like ground truth, labelling noise, boundary box and various types of annotations and segmentations.
There is only one show left in this season of Reasons to do AI with Friends. Make sure to tune in next Wednesday, February 19, at 19:30 CET for the season finale. The final episode, “The one with whole body MRI”, will be hosted by Andrea Rockall, from the Imperial College London, together with her colleagues Ben Glocker and Theo Barefoot, and will be broadcasted live from London, England. Registration is free of charge.
All episodes will be available on-demand for purchase on connect.myesr.org.
Read more on Episode 1, “The one where we classify breast cancer with MRI” here.
Read more on Episode 2, “The one where we tackle fibrosis” here.
Read more on Episode 3, “The one with the sixpack” here.
Read more on Episode 4, “The one that might be renal cancer” here.
Want to deepen your AI knowledge? Check out the reading recommendations from our experts here.