AI & ML · DeepTech
When to trust the scan that never happened?
Hosted by Machine Learning in Healthcare Berlin
Tuesday 1 December 2026 · 6:00pm · Berlin Institute of Health (BIH) · hosted by Machine Learning in Healthcare Berlin
about this room
A talk on judging synthetic medical images: CT and MRI scans that generative models produce but nobody ever acquired.
Adam Hilbert, AI scientist in neuroradiology and stroke research at Charité, uses his recent papers to show that looking realistic, matching the individual patient and being useful for a downstream task are three different properties that often disagree, and that neither standard metrics nor expert raters reliably catch them. The audience gets to test its own eye on examples. Doors 18:00 with food, talk at 19:00, networking after, in the Atrium on the 5th floor of the Spreepalais.
Free; Meetup RSVP is request-to-join. Sponsored by the QUEST Centre at BIH.
a concrete, evidence-led answer to a question generative AI is forcing on medicine. Best for ML engineers and researchers in medical imaging and health AI.
Adam Hilbert · AI Scientist, Charité – Universitätsmedizin Berlin
getting there
rooms like this, every friday morning.
more rooms this week
Digital Health Conference 2026 (#dhc26)Tue 24 Nov · 9:00am · paid
Thursday Gathering #80 — Asia Berlin Satellite EventThu 26 Nov · 5:00pm · free
Digital Health Conference 2026 (#dhc26)Tue 24 Nov · 9:00am · paid