💥Exceptionally grateful to the @HelmsleyTrust for continued support of the #STARConsortium with a 2.7 Mio. USD grant for developing a novel AI based imaging approach for a 'virtual' full thickness intestinal biopsy in stricturing Crohn's disease 💥https://t.co/tXUYOkAJuK
Love to round out the year listening to close friends and collaborators @Pallavi_Tiwari@BrennanFlan and Tom Desilvio talk about topics around #imaging and #AI; of deep interest to all of us.
Enjoy!
🚨A new episode of the @radiology_rsna: Imaging Cancer podcast is now available!
@BrennanFlan & Thomas DeSilvio talk with @Pallavi_Tiwari about how AI is shaping individualized imaging and real-world clinical care. Enjoy!
https://t.co/s5E2kManB7
🔎Fusion enhanced biological interpretability. Intelligently fused FMs produce:
✔Cleaner and more localized attention maps
✔Better separation of tumor vs. benign tissue in unsupervised clustering
➡️ Smarter integration, not bigger models, leads to better pathology #AI.
#NewPaper from our group on #AI and #digitalpathology, led by Dr Brennan Flannery!
We asked a key question: How similar are these models really, and can we intelligently combine them to do better than any single model alone?
https://t.co/D25Q01aDX0
🔎 We propose an “intelligent fusion” strategy that removes redundant FM features while preserving complementary info.
➡️ Across diseases, this approach:
✔ Improved F1 and AUC scores vs. individual FMs
✔ Increased attention to true tumor regions while reducing spurious focus
Fascinating and exciting sessions this morning at the 3rd @Emory_AIHealth symposium showcasing advances in real-time clinical #AI, lessons from industry toward #translation, and a keynote on #AgenticAI for scientific discovery @EmoryUniversity
Delighted to share that the @HelmsleyTrust has awarded a $3.2 million grant to the three of us: myself @WashUMedGastro@WashUDeptMed, Satish Viswanath, PhD @satishev, an associate professor in the Departments of Pediatrics & Biomedical Engineering @EmoryPediatrics@CoulterBME & David H. Ballard, MD @DavidBallardMD@WashUMedMIR.
The project will validate:
👉Newly developed radiologic definitions for fistula improvement and response through the TOpCLASS fistula consortium. https://t.co/9PZOLbctcy
👉 Fully automated fistula segmentation neural network model on routine pelvic MRIs.
https://t.co/UG0vNQXVj9
👉Further, using advanced computational approaches and artificial intelligence (deep learning) methods, the grant would allow the development of novel radiomics tools to measure fistula response that can be eventually deployed in clinical practice and trials in perianal Crohn’s disease drug development and surgical care pathways.
Don't forget to tune in tomorrow, October 10th as we welcome @satishev for a session on "The Case of the Hidden Signatures: Designing Imaging AI to Bridge Patterns, Predictions & Precision Medicine."
Learn more: https://t.co/TOhgZklqq3
Our ML Healthcare is looking forward to hosting @satishev as he explores how handcrafted AI tools and computational imaging features can integrate diverse data modalities to predict treatment response and patient outcomes, enabling precision medicine in cancers and other diseases. 🏥
Thanks to @LeemaMurali, @AnasZaf79138457 and @oumayma_es_ for organizing this session!
Learn more an join us on October 10th: https://t.co/vGBiMJchkD
Very excited to collaborate with Maneesh Dave @UCDavisHealth on this novel effort funded by @NIBIBgov to develop #AI and #DigitalPathology approaches toward investigating biological and mechanistic drivers of #IBD and #GvHD https://t.co/TLaWbfoOiy