A Review Of The Machine Learning Datasets In Mammography, Their Adherence To The Fair Principles And The Outlook For The Future

This article discusses the use of Bishop & Hanks’ framework to evaluate the compliance of datasets with the FAIR principles. Eight key datasets were identified and carefully selected based on their relevance to the research topic and the availability of comprehensive mammographic imaging data within them. The Digital Database for Screening Mammography (DDSM) is a popular large-scale mammographic dataset released in 1996 and has been cited in more than 80 distinct papers in mammographic machine learning and artificial intelligence. The dataset consists of 2,890 cases, including left and right cranio-caudal (CC) and mediolateral-oblique (MLO) views, for a total of 11,560 mammographic images.