Radiology AI is advancing quickly, if this year's annual Society of Imaging Informatics in Medicine Conference on Artificial Intelligence in Medical Imaging (SIIM-CAIMI) meeting is any indication.
Attendees benefited from the opportunity to see where AI is heading in clinical practice, including scientific sessions that focused on AI’s use in assisting with disease diagnosis and with clinical operation efficiency.
One study that took third place in the category of Best Research Abstract – Oral Presentation explored how a large language model (LLM) approach can automate the finding of critical results and monitoring turnaround times in pediatric radiology.
Radiology AI is advancing quickly, if this year's annual Society of Imaging Informatics in Medicine Conference on Artificial Intelligence in Medical Imaging (SIIM-CAIMI) meeting is any indication.
Attendees benefited from the opportunity to see where AI is heading in clinical practice, including scientific sessions that focused on AI’s use in assisting with disease diagnosis and with clinical operation efficiency.
One study that took third place in the category of Best Research Abstract – Oral Presentation explored how a large language model (LLM) approach can automate the finding of critical results and monitoring turnaround times in pediatric radiology.
Data scientist Pooja Pathak from Stanford Health told AuntMinnie that this approach can continuously monitor every finalized report in-house, with no need for commercial tooling or external data transfer.
Pooja Pathak discusses her team's AI pediatric imaging pipeline and how it could improve turnaround times.
Keynote presentations also showed how radiology should not be siloed off from other specialties in using AI for disease diagnosis and care optimization.
In his talk, Faisal Mahmood, PhD, from Harvard Medical School and Brigham and Women's Hospital, shared early results from Apollo, a multimodal temporal foundation model. The model has been trained and evaluated on approximately 25 billion records from 7.2 million patients in the Mass General Brigham health system.
This data represents 28 distinct medical modalities and 12 major medical specialties. Apollo takes data from radiology, pathology, genomics, and clinical data to create a single, temporally aligned representation of a patient’s condition.
And in another keynote talk, Nabile Safdar, MD, from Emory Healthcare, outlined what’s currently working in medical imaging AI and how the alignment of regulatory, financial, and cultural change along with the technology will shape AI’s future in healthcare.
Nabile Safdar, MD, shares what is currently working in medical imaging AI and what could be done better in clinical practice.
Another featured session focused on how integrating imaging and clinical context, reason across complex tasks, and clinician support are needed beyond benchmarking, given by Pranav Rajpurkar, PhD, from Harvard University. Sessions also provided talks toward research in agentic AI and vibe coding.
SIIM-CAMI leaders also shared their thoughts on AI’s future in radiology, including what needs to be done in the next step of the technology’s evolution. And program chairs spoke with AuntMinnie about what they believe will be seen more at SIIM-CAIMI 2027, as well as why there should be optimism toward AI in medical imaging.
CAIMI program co-chair Peter Chang, MD, says multidisciplinary collaboration will be needed to advance AI in healthcare.