2020/04/09 by Mark Halling‐Brown, Lucy M. Warren, Halling-Brown, Mark D +17 · 4 citations
Computer Science · Medicine · #AI in cancer detection #Radiomics and Machine Learning in Medical Imaging #Digital Radiography and Breast Imaging
paper · pdf · doi:10.48550/arxiv.2004.04742
A major barrier to medical imaging research and in particular the development\nof artificial intelligence (AI) is a lack of large databases of medical images\nwhich share images with other researchers. Without such databases it is not\npossible to train generalisable AI algorithms, and large amounts of time and\nfunding is spent collecting smaller datasets at individual research centres.\nThe OPTIMAM image database (OMI-DB) has been developed to overcome these\nbarriers. OMI-DB consists of several relational databases and cloud storage\nsystems, containing mammography images and associated clinical and pathological\ninformation. The database contains over 2.5 million images from 173,319 women\ncollected from three UK breast screening centres. This includes 154,832 women\nwith normal breasts, 6909 women with benign findings, 9690 women with\nscreen-detected cancers and 1888 women with interval cancers. Collection is\non-going and all women are followed-up and their clinical status updated\naccording to subsequent screening episodes. The availability of prior screening\nmammograms and interval cancers is a vital resource for AI development. Data\nfrom OMI-DB has been shared with over 30 research groups and companies, since\n2014. This progressive approach has been possible through sharing agreements\nbetween the funder and approved academic and commercial research groups. A\nresearch dataset such as the OMI-DB provides a powerful resource for research.\n