Visual adaptation and the perception of medical images – Professor Michael Webster, Foundation Professor of Psychology, University of Nevada, USA
In this talk delegates will learn:
• How and why sensory systems continuously adapt their properties.
• How this adaptation adjusts to physical differences between environments and physiological differences between individuals.
• That adaptation affects all perceptual judgments and thus every visual experience.
• Why these adjustments are important for understanding the perception and interpretation of medical images.
Michael Webster is Foundation Professor of Psychology at the University of Nevada, Reno, and Director of UNR’s Center for Integrative Neuroscience. He received his doctorate from UC Berkeley in 1988 and was a postdoctoral fellow at Cambridge University until 1994. Dr. Webster is a vision scientist who studies colour and form perception and how visual processing adapts to changes in the environment or the observer. He has published widely on both adaptation and individual differences in perception, and in 2019 was awarded the Verriest Medal from the International Colour Vision Society for his contributions to understanding human colour vision.
How to assess AI for screening – Professor Nico Karssemeijer, Founder of ScreenPoint and Professor of Radiology, Radboud University, Nijmegen, The Netherlands
Professor Nico Karssemeijer is professor of Computer Aided Diagnosis (CAD) at the Radboud University Nijmegen and CEO and co-founder and CEO of ScreenPoint Medical. He studied physics at Delft University of Technology and graduated at the Radboud University Nijmegen in 1989. The research group he formed at Radboud University Medical Centre gained worldwide recognition for its pioneering work in breast imaging and computer aided diagnosis. He served as associate editor of IEEE Transactions on Medical Imaging, was member of the executive editorial board of Physics in Medicine and Biology, and member of the editorial boards of Medical Image Analysis and Journal of Medical Imaging. He also served as symposium chair of SPIE Medical Imaging and was member of the scientific program committee of RSNA.
From 1996 to 2008 Nico Karssemeijer was consultant of the Silicon Valley based company R2 Technology and made major contributions to the ImageChecker, which was for two decades the most widely used CAD system in radiology. He was principal investigator in the EU funded projects SCREEN and SCREEN-TRIAL, which had a strong impact on digitisation of breast screening programs and was coordinator of the H2020 project ASSURE on technology development for personalised breast screening.
In 2014 Nico Karssemeijer founded ScreenPoint Medical BV in Nijmegen, The Netherlands, which develops AI solutions for breast imaging. He is also co-founder of Volpara Health Technologies (Wellington, NZ), a company that provides solutions for quality and workflow improvement, and QView Medical Inc. (Los Altos, CA, USA), which develops AI for automated breast ultrasound.
Evaluation processes for AI – Professor Ken Young, Head of Research, NCCPM
This talk will help delegates to understand:
• Types of evaluation
– as 2nd Reader
– decision support
• Dataset selection for retrospective testing of AI as a 2nd reader.
• Practical issues when testing AI.
• How to compare human readers with AI tools.
Professor Ken Young led the National Coordinating Centre for the Physics of Mammography (NCCPM) from 1990 to 2019. He currently leads research at NCCPM, and is a Visiting Professor of Medical Physics at the University of Surrey. He led the OPTIMAM research projects funded by Cancer Research UK which created the OPTIMAM Image Database; one of the world’s largest mammography image databases, which is shared with leading academic research groups and commercial companies. Collaboration in the development and evaluation of artificial intelligence techniques to automate breast cancer detection is a major focus of his current research.
Developing AI for breast density, risk assessment and generalisability of models – Dr Lucy Warren, Clinical Scientist, Royal Surrey NHS Foundation Trust
This talk will cover:
• Generalisability of artificial intelligence in breast imaging.
• Use of AI to predict breast density.
• Use of AI to predict breast cancer risk.
Dr Lucy Warren has worked at the Royal Surrey County Hospital for almost 10 years. She has completed a collaborative PhD with the University of Surrey and is registered as a Clinical Scientist, in the field of Diagnostic Radiology and Radiation Protection. Lucy’s research interests include simulation of breast cancers, observer performance studies to optimise breast imaging, deep learning and modelling outcomes of breast cancer screening.
She also lectures on the physics of breast imaging and image quality evaluation on the Medical Physics MSc course at the University of Surrey.
How might technology lead to quality improvements in the service – Dr Melissa Hill, Consultant Imaging Scientist, Volpara Solutions Ltd
This talk aims to:
• Describe potential benefits of automated analysis of breast positioning and compression in screening mammography.
• Describe areas where automated algorithms could help the NHSBSP achieve image quality performance targets.
• Review causes of technical repeat (TP) and recall (TC) in the NHSBSP and how automated algorithms could help reduce TP and TC numbers.
Melissa Hill has worked in the field of mammographic imaging physics for 15 years, with applications including digital mammography (DM), digital breast tomosynthesis (DBT), dedicated breast CT, and contrast-enhanced versions of both DM and DBT. A native of Canada, Melissa now lives and works in France as a Consultant Imaging Scientist to Volpara Solutions Ltd. Her role includes image analysis for quantitative breast density measurement, analysis of big-data, breast dosimetry, and development of novel techniques for image quality measurement. Melissa works at the interface of industry and academia and enjoys collaborating with researchers globally.