Medical Imaging 2010: Physics of Medical Imaging

Breast density has been positively linked to the risk of developing breast cancer. We have developed a stepwedge-based method that provides an estimate of the amount of glandular tissue both by volume and by area. The method has been applied to the mammograms of 1,289 women in the UK breast screening programme. The mammograms of 29 women (116 images) were analysed by three independent operators in order to assess inter-observer variability of the method. From this sample, a subset of 24 images were analysed on 10 separate occasions by the same operator to assess intra-observer variability. Additionally, 168 images were analysed using the stepwedge method and assessed by tworadiologists who independently scored percentage density using a visual analogue scale.There was little intra-observer variability, with average coefficients of variation ranging from 3.49 to 5.73. However, there were significant differences in the volumes of glandular tissue obtained by the three operators. On further inspection, this was found to be caused by variation in the way in which the operators marked the breast edge, rather than by their definition of the other landmarks in theimage or by variation in the measured breast thickness. There was good correlation between percentage of dense area of the breast and radiologists’assessment of percentage density, particularly for dense and fatty breasts. A similar pattern of agreement was found between the two radiologists. Based on analysis of thresholds used in the stepwedge method, radiologists’ definition of a dense pixel is one in which the percentage of glandular tissue is between 10 and 20% of the total thickness of tissue.