Artifical intelligence: Impact on mammogram interpretation
02 Nov 2017
Deep learning, a form of machine learning, has recently received a substantial amount of attention in the radiologic community. This talk will briefly review the basics of machine learning and (deep) artificial neural networks.
It will then bridge to the latest clinical research from their group: to estimate the ability of current state-of-the-art algorithms, a head-to-head comparison of the cancer detection rate of radiologists versus a generic deep learning image analysis software was performed. The overall performance was comparable, whereas the software exhibited consistently higher sensitivity but lower specificity [1].
Lastly, the limitations of machine learning and the opportunities for radiology in the near future will be discussed.
Educational aims:
• To understand the basic concepts behind machine learning and deep neural networks
• To understand the limitations of those methods
• To summarise the current state of machine learning research in breast imaging
References:
1. Becker A, Marcon M, Ghafoor S, Wurnig M, Frauenfelder T, Boss A. Deep Learning in Mammography. Investigative Radiology. 2017;52(7):434-440.
Further reading:
• Wang S, Summers R. Machine learning and radiology. Medical Image Analysis. 2012;16(5):933-951.
0.5 CPD credit
Watch the video and complete the online self-reflection form. Go to "My events" to download your certificate. You must login to watch the video and receive CPD.
It will then bridge to the latest clinical research from their group: to estimate the ability of current state-of-the-art algorithms, a head-to-head comparison of the cancer detection rate of radiologists versus a generic deep learning image analysis software was performed. The overall performance was comparable, whereas the software exhibited consistently higher sensitivity but lower specificity [1].
Lastly, the limitations of machine learning and the opportunities for radiology in the near future will be discussed.
Educational aims:
• To understand the basic concepts behind machine learning and deep neural networks
• To understand the limitations of those methods
• To summarise the current state of machine learning research in breast imaging
References:
1. Becker A, Marcon M, Ghafoor S, Wurnig M, Frauenfelder T, Boss A. Deep Learning in Mammography. Investigative Radiology. 2017;52(7):434-440.
Further reading:
• Wang S, Summers R. Machine learning and radiology. Medical Image Analysis. 2012;16(5):933-951.
0.5 CPD credit
Watch the video and complete the online self-reflection form. Go to "My events" to download your certificate. You must login to watch the video and receive CPD.
Duration:18 mins
Speaker info
Anton Becker
Dr Anton Becker Research Fellow, University Hospital of Zurich Dr Anton Becker holds a medical degree from the University of Zurich. He initially developed an interest in artificial neural networks while working on his Master’s thesis in neuroanatomy under the supervision of Professor H.P. Lipp and Dr I Amrein in 2009. After starting his training in diagnostic and interventional radiology at the University Hospital of Zurich, he completed a research fellowship with Professor Andreas Boss focusing on MRI and machine learning. He has a strong interest in functional, metabolic and oncological imaging as well as machine learning applications in radiology.