Matjaž Jogan

New work from @PCASOLab led by @Qiu84046087 now on @arxiv: Detection and segmentation of transient objects in video using only sparse object presence labels, developed to automate annotation of surgical videos #wsvos #wsvod #wsvostop #surgicalAI

Congrats to @pennsurgery @GRASPlab @PCASOLab postdoc @Qiu84046087 on his #WACV2025 presentation on weakly supervised learning for surgical video analysis! Great collaboration w/ @LifelongML_Penn Coauthors in lab include @matjogan @SKoushik

How do we judge the viscosity of materials? It turns out it depends not only on how they move, but also on how they look. Our latest work with the incredible Jeffrey Martin out now royalsocietypublishing.org/doi/10.1098/rs…

Happy holidays from the PCASO Lab team! Thanks to everyone who made it to breakfast and missing those already out of town!

Finally found time to update my webpage with more details on our recent open-source dataset releases: a relabeled version of MedQA and a differential diagnosis dataset in dermatology. Great examples of dealing with rater disagreement for evaluating AIs: davidstutz.de/wMDSi

More training data isn’t always better! We (joint work w.@rajiinio and @irenetrampoline) will present our new work: The Data Addition Dilemma, at MLHC 2024 this week! (1/5)

New publication from @PCASOLab in Global Surgical Education @GSE_JASE where we discuss data quality, bias and uncertainty in surgical AI link.springer.com/10.1007/s44186… @pennsurgery #surgicalAI #surgicaldatascience

Quality of training data is known to dictate quality of #MachineLearning in healthcare. But what is data quality in this context? The METRIC-framework serves as a basis for systematically assessing medical training data as prerequisite for trustworthy ML. nature.com/articles/s4174…

Come compete: cvs-challenge.grand-challenge.org! #miccai2024 @MICCAI_Society @Laparoscopes @p_masca @dpkalapatt @CammaLab

Can you spot #AI generated surgical images? Surgeons please help us evaluate a novel surgical text-to-image gen model by completing a short survey 🧑‍⚕️👩‍⚕️ Survey 👉 t.ly/endogen ArXiv 👉 arxiv.org/abs/2407.09230 #SurgicalDataScience #GenAI #medtwitter #SoMe4Surgery

1/ How are #ChatGPT and other #LLMs used in #Medicine? And what are the ethical implications? We say: It’s a social experiment unfolding right in front of us! Check out @robranisch ‘s and my new paper in @npjDigitalMed nature.com/articles/s4174… Let me give you a quick tour

Looking for a new paradigm for your next research project? Then look no further: nature.com/articles/s4156… (1/6)

Our comprehensive review of #AI in surgery is just out @NatureMedicine nature.com/articles/s4159… an inspiration and joy to work with super-talented @chrisvarghese98 @ogrady_greg @ewenharrison on it Free access rdcu.be/dHL7g

.@pennsurgery @PennEngineers students and researchers from @PCASOLab presented work at the 2nd annual @PennMedicine Naji Surgery Research Day! Great work spanning clinical outcomes in GI surgery, #computervision for thoracic surgery, digital literacy, & efficient algorithms

Our 12 scaling laws (for LLM knowledge capacity) are out: arxiv.org/abs/2404.05405. Took me 4mos to submit 50,000 jobs; took Meta 1mo for legal review; FAIR sponsored 4,200,000 GPU hrs. Hope this is a new direction to study scaling laws + help practitioners make informed decisions

2 papers out @naturemethods on metrics for #AI image analysis including in surgical images w/ @GRASPlab @PCASOLab's @Laparoscopes as coauthor   nature.com/articles/s4159…   nature.com/articles/s4159…   #MetricsReloaded #surgicalAI

🍻EXCITING DAY: Our @naturemethods sister publications on #MetricsReloaded are out🎉. Recommendations: rdcu.be/dysg5 Pitfalls: rdcu.be/dysgX @helmholtz_image @bias_sig @ProjectMONAI @NCT_HD @DKFZ Important for the community: 1/

We made a free, open course on neuroscience for people with a machine learning / quantitative background. Get deep in computational neuroscience, or just indulge your neurocuriousity. I believe that neuro and ML can learn from each other and do better together than on their own.

1/2 📣New #dataset announcement! #Endoscapes 2023 includes 201 laparoscopic videos annotated for scene segmentation, object detection, and critical view of safety assessment. arXiv: arxiv.org/abs/2312.12429 Code: github.com/CAMMA-public/E… #SurgicalDataScience #SurgicalAI

New in @AcadMedJournal from @PCASOLab's @Laparoscopes & Measey scholar @shubha_vasisht: a review of #AI in #MedEd that highlights key challenges & important opportunities. journals.lww.com/academicmedici…