Speakers - Annual Scientific Meeting
Fiona Gilbert
Fiona J. Gilbert, FRCR, FRCPS,
FRCP, FACR, FRSE, FMedSci, is Professor of Radiology at the University of
Cambridge, UK. Her research is focused on imaging techniques relating to breast
cancer and oncology. Previously she has
evaluated digital breast tomosynthesis and computer aided detection in the
breast screening programme. She undertakes research in risk adapted stratified
breast screening using Abbreviated MRI, Tomosynthesis, Whole Breast Ultrasound
and Contrast Enhanced Mammography. She uses multimodal functional imaging with
MRI and PET to examine the tumour environment using breast cancer as a model
and correlating this with the tumour molecular profile. She
evaluates new imaging technology and is currently working on Artificial
Intelligence in Imaging.
Since 2012 Professor Gilbert has
been awarded fifteen competitive grants worth over £20M. She was a member of grant
funding panels - NIHR HTA Board, the EME Board and is on a number of advisory
panels. She was a previous associate editor of Clinical Radiology.
Professor Gilbert has over 260
peer reviewed publications, 5 book chapters and numerous conference abstracts. She is a regular speaker at international
Radiology conferences in Chicago and Vienna and was awarded Honorary membership
of Radiological Society of North America in 2019, Honorary fellowship of the
American College of Radiologists, the Gold Medal from the European Society of
Radiology and fellowship of the Royal Society of Edinburgh in 2021 and
fellowship of the Academy of Medical Sciences. She is immediate past President
of the European Society of Breast Imaging.
Title: Prospective trials of AI in screening in the UK and internationally
Published retrospective studies have shown that AI can match the performance of a single reader with some non-inferior to the standard double reading. Several countries have started prospective studies to confirm these findings and to test whether one reader can be replaced safely. Interim results from the Swedish MASAI trial has shown that a single reader using AI can increase cancer detection (from 5/1000 to 6/1000) while maintaining low recall rates with a marked saving in reader time compared to double reading. The Swedish ScreenTrust CAD paired design study of AI independently working with two independent readers has shown similar results. The Danish Capital Region breast cancer screening programme has implemented AI in clinical practice with ongoing evaluation (MAGIC – MammGraphy AI in Breast cancer diagnostics). An AI score 1-5 is single-read, AI score 6-10 requires double-reading, and where the AI score is >9.98 the woman will be recalled to assessment based on AI only. The PRAIM Study (PRospective multicentre observational study of an integrated Artificial Intelligence (AI) system with live Monitoring) prospectively investigates double-reading versus single-reading + AI algorithm, within a cohort of 400,000 women (age 50-69 years) in Germany. Norway is setting up a single reader with AI study. The AITIC trial (Artificial Intelligence in Breast Cancer Screening Programs in Córdoba) involving 27,000 women will investigate double-reading versus reading strategy based on the AI score <8 (low probability of cancer) will not be evaluated by any radiologist, score >7 double-reading. The AI-STREAM (Artificial Intelligence for breast cancer screening in mAMmography) prospective trial investigates single-reading versus single-reading + AI algorithm (computer aided detection/diagnosis) in 32,714 women in Korea reporting diagnostic accuracy with and without CAD prompts. There are several studies either in set up and ongoing in the USA, some with mammography and others with DBT which will be reviewed at this meeting.
British Society of Breast Radiology is supported by:
Supported by:
Managed and payment processed by: KC Jones conference&events Ltd
Telephone: 01332 227773
Email: bsbr@kc-jones.co.uk