@ucsfci2.bsky.social
The Center for Intelligent Imaging (Ci2) is focused on applications of artificial intelligence and image analysis tools in medical imaging.
@ucsfci2.bsky.socialA new comparative study led by @ucsfci2.bsky.social's Dr. Alexandra Gersing found that deep learning-based denoising of low-field 0.55T knee MRI improved diagnostic accuracy from 0.83 to 0.98, achieving results comparable to conventional 3T. www.doi.org/10.1016/j.ej...
@ucsfci2.bsky.socialA new study in #npjDigitalMed by @ucsfci2.bsky.social's @drdremdphd.bsky.social presents a #DeepLearning framework that generates uncertainty estimates for meningioma segmentation on brain MRI, supporting safer clinical AI deployment. bit.ly/47sqBOf
@ucsfci2.bsky.socialNew narrative review from @ucsfci2.bsky.social's Dr. Sharmila Majumdar & her former student Dr. Katharina Ziegeler examines how AI can opportunistically assess bone mineral density & bone microarchitecture from routinely acquired imaging across radiography, CT & MRI. bit.ly/4hAf3Pa
@ucsfci2.bsky.socialFinding a deep pediatric brain tumor's subtype usually means a biopsy. Emma Duprat tested whether AI uncertainty could flag it noninvasively. Louan Bardou tackled detecting white matter lesions in adolescents without FLAIR. Learn more about their research: bit.ly/4719oeC
@ucsfci2.bsky.socialMeet the #postdoc making sense of breast cancer MRIs from more than 30 sites. Pouya Metanat studies how scanner & software variability affects the reading of treatment response, so care can be personalized. Happy #PostdocAppreciationWeek! #UCSFProud
@ucsfci2.bsky.socialWe're recognizing Atlas Haddadi Avval this #PostdocAppreciationWeek. Her research uses #AI to compare a patient's MRI scans over time & quantify how a pediatric brain tumor is changing. The aim is a more objective, reproducible way to track disease & inform treatment. #UCSFProud
@ucsfci2.bsky.socialMost tools that segment white matter lesions rely on FLAIR imaging. Louan Bardou reached comparable accuracy without it. Emma Duprat used AI segmentation uncertainty to read pediatric brain tumor subtypes. Learn more: bit.ly/4719oeC
@ucsfci2.bsky.socialBuilding tools to read prostate MRIs takes data with reliable answers attached. @ucsfci2.bsky.social's UCSF-ProstateMR dataset pairs 973 prostate scans with the biopsy results that show what was truly there, giving researchers a solid basis to test their work. bit.ly/4zvkp5d
