IntroductionIn Western countries, about one in eight to ten women develops breast cancer during their lifetime [ 1 ]. Well-confi rmed risk factors for breast cancer are reproductive factors (e.g., non-parity, late fi rst birth, early menarche, and late menopause) [ 2 ], hormone replacement therapy (HRT) [ 2 ], genetic factors [ 3 ], ionizing radiation [ 4 ], and high breast density on mammography [ 5 ]. There is also evidence that lifestyle factors can increase the risk, such as high alcohol consumption [ 6 ] and low physical activity [ 7 ]. Mammography plays a central role in early detection, since it can show changes in the breast before the patient or a physician can feel them. Because of its clinical effectiveness, the technique has been used for detection of breast cancer for more than half a century [ 8 ], and since several decades also for screening of asymptomatic women. Screening mammography is one of the largest public health efforts to promote women's health, starting with pilot studies and proceeding with larger randomized controlled trials (RCTs) to determine the potential benefi ts [ 9 -16 ]. In randomized controlled trials (RCTs), mammography screening was shown to reduce the breast cancer mortality by approximately 20-30 % [ 17 , 18 ], which has led to today's population-based mammography screening programs. Screen-fi lm mammography (SFM) has since long been the standard technique in breast cancer screening but advances in the digital detector technology and computers have paved the way for digital mammography (DM), and since the US Food and Drug Administration's (FDA) approval of the fi rst commercial systems in the year 2000, two-dimensional (2D) DM became the accepted standard of care in breast cancer screening and diagnosis in North America as well as in Europe and in a majority of other developed countries. The use of DM has increased rapidly since it offers many advantages compared to SFM, such as higher image quality and/or lower radiation dose to the breast, omitting recalls due to technical failure, increased patient throughput, more effi cient image management, and telemammography. However, despite the improvements, the mammographic accuracy has shown to be imperfect, and reader variability that may be infl uenced by various factors such as radiologist experience, case diffi culty, and varying practices at different mammography centers has remained a great challenge. Sensitivities have been estimated from 68 % (as low as 48 % for very dense breasts) to 88 % at specifi cities from 82 to 98 %, suggesting that further improvements can be made [ 19 , 20 ]. A major problem lies in the nature of the two-dimensional (2D) technique. Because a conventional mammogram is a 2D projection of the breast onto the detector plane, over-projected normal tissue (anatomical noise) can restrain the accuracy of mammography. In the clinical practice, cancer detection may be limited, particularly in younger women and in those with a high parenchymal density since the mammographic evidence of the tumor may be co...