Asst. Professor (Radiology) @Stanford, prev. @Unispital_USZ | MD, MSc | prev. ML Postdoc @StanfordAIMI | tweets my own, RT ≠ endorsement
Palo Alto, CA
Joined January 2017
- Tweets662
- Following263
- Followers1.6K
- Likes3.1K
Pinned Tweet
Our review on Agentic Systems in Radiology was just published in @radiology_rsna ! We explore agent fundamentals, agentic topologies, challenges, evaluation, and how these concepts apply to radiology as an environment. This is a go-to resource for those who want to familiarize themselves with the topic.
Glad that although the speed of developments exceeds peer review, most presented concepts still hold :) Harness engineering and RSI are more important now, MCP is more broadly accepted by now (lol), and a few new benchmarks and eval strategies have emerged.
Christian Bluethgen retweeted
🤖 Radiology AI is moving beyond tasks toward orchestrating complex workflows.
This review explores LLM-driven agentic systems, potential radiology applications, evaluation strategies & challenges including error cascades, tool use, and IT integration.
pubs.rsna.org/doi/10.1148/ra…
Christian Bluethgen retweeted
LLM-driven agentic systems show promise for complex radiology tasks but require careful integration, multilayered evaluation & attention to key challenges. @cxbln @jnkath @Michael_D_Moor @Dr_ASChaudhari @curtlanglotz @farhadnz bit.ly/4imspNM
where else would we have learned how to parse HTML with regex stackoverflow.com/a/1732454
Some thoughts I shared in yesterday's AIMI Grand Rounds talk I personally found to be useful re:LLM-assisted coding
@StanfordAIMI
Christian Bluethgen retweeted
Introducing CheXthought, Stanford AIMI’s newest dataset for AI reasoning in chest X-rays. 103,592 reasoning traces and 6M+ spatial annotations across 50,312 X-rays, contributed by 501 radiologists in 71 countries.
🔗 Download the dataset: bit.ly/CheXthought
#StanfordAIMI
🙌 Huge thanks to my brilliant coauthors from the University of Zurich, ETH Zurich, RWTH Aachen, TU Dresden, and Stanford: @dave_van_veen @DanielTruhn @jnkath @Michael_D_Moor @MPolacin @Dr_ASChaudhari Thomas Frauenfelder @curtlanglotz @krauthammerlab @farhadnz @StanfordAIMI @StanfordRad @Stanford @Unispital_USZ @UZH_en @ETH_AI_Center @ETH
Christian Bluethgen retweeted
Please share widely!
I’m recruiting PhD students @UMBME for Fall 2027.
Interested in applying AI to multimodal data (medical imaging, biomechanics, human movement) to tackle clinical challenges in osteoarthritis, orthopaedics, lung cancer & beyond? Please apply by Dec 1.
Christian Bluethgen retweeted
Hi all, I have a new result I'm excited to share. It turns out that *all existing implementations* of softmax and cross-entropy (that's JAX, PyTorch, textbooks, etc) have a subtle bug --- two bugs, actually. A simple rewrite fixes both, and eliminates "slingshots" when grokking:
reminds me of @polynoamial's example of airgapped machines communicating through heat
hmm
My brother made a fun website: deprecation.org
Enter your job and it tells you when you'll be deprecated, and exactly how. For me it's... pretty soon and pretty accurate!
He's a very creative person (artist turned frontend dev to pay bills, now obv worried by AI) and this is full of little nuggets and easter eggs. It's an eerie mix of fun and jokes and real worries. Check it out!
Christian Bluethgen retweeted
Just watched the @netflix Instadocs about the @OpenAI cyberattack on @huggingface. Super disappointing, as they went for maximum anthropomorphism of the technology and full doomer mode.
I spent more than two hours with them trying to correct all their mistakes and ground this in reality, explaining that agents are technological systems, not aliens. But it looks like the drive to scare people for maximum views was too strong.
Also, they didn’t mention the lack of containment and monitoring of the agents, how closed-model safeguards blocked our defensive work, or how we defended ourselves with AI using an open model. And that we still lack a lot of information and transparency, including full agent traces and that companies need to remain accountable for what their agents do.
It ends up being a fun sci-fi movie though!
I see at least 4 errors
Just asked Claude to do the labelling, and it did it perfectly. Amazing!
I fear our prefrontal cortices will become redundant in the near future 💀
#MedTwitter
Christian Bluethgen retweeted
There's still time to share your work at Stanford #AIPlusHealth2026. Submit an abstract for a 5-minute lightning talk at our virtual conference, Dec 8–9.
Deadline: Oct 15, 11:59pm PT
🔗 stanford.io/4s1zs37
#HealthcareAI #StanfordAIMI
Abstract submissions open for Stanford AI+HEALTH 2026, online Dec 8–9. Share your latest research at one of Stanford’s largest meetings focused on AI in health and medicine. Selected abstracts will be presented as 5-min lightning talks.
🕓Submit by Oct 15: bit.ly/4iCmdmQ
Christian Bluethgen retweeted
DINOv2 has never seen a caption, and Qwen3 has never seen an image. We still aligned their embedding spaces without a single image-caption pair.
It even works when the images and the captions come from different datasets.
Project page: dominik-schnaus.github.io/un…
⬇️
Christian Bluethgen retweeted
Congratulations to AIMI affiliated faculty member Sergios Gatidis, MD, on receiving the @NIH Director’s New Innovator Award! His research aims to advance medical imaging AI beyond pattern recognition toward understanding anatomy and physiology. 🔗 stanford.io/4B1MbXM
