BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Multimodal AI for Radiology Applications - Kenza Bouzid\, Applied 
 Researcher at Microsoft Health Futures 
DTSTART:20241121T130000Z
DTEND:20241121T140000Z
UID:TALK224464@talks.cam.ac.uk
CONTACT:116317
DESCRIPTION:Radiology reporting is a complex task requiring detailed medic
 al image understanding and precise language generation\, for which generat
 ive multimodal models offer a promising solution. However\, to impact clin
 ical practice\, models must achieve a high level of both verifiable perfor
 mance and utility. We augment the utility of automated report generation b
 y incorporating localisation of individual findings on the image– a task
  we call grounded report generation– and enhance performance by incorpor
 ating realistic reporting context as inputs. We design a novel evaluation 
 framework (RadFact) leveraging the log ical inference capabilities of larg
 e language models (LLMs) to quantify report correctness and completeness a
 t the level of individual sentences\, while supporting the new task of gro
 unded reporting. We develop MAIRA-2\, a large radiology-specific multimoda
 l model designed to generate chest X-ray reports with and without groundin
 g. MAIRA-2 achieves state of the art on existing report generation benchma
 rks and establishes the novel task of grounded report generation.\n \n
LOCATION:Computer Laboratory\, William Gates Building\, LT2
END:VEVENT
END:VCALENDAR