Impact of prior mammograms on observer performance in screening mammography

Abstract

Aim: Many screening programs store screening mammograms for future reference. This thesis examined the influence of prior mammograms on the efficacy of screening mammography interpretation.

Methods: The three original studies were conducted using cases from the BreastScreen Reader Assessment Strategy (BREAST) platform. In the first study, eight radiologists independently interpreted 72 screening mammograms with and without prior mammograms in two reading sessions. In the second study, 13 radiographers interpreted two sets (“Prior” and “No prior”) of mammograms. Reader performances with prior mammograms versus without prior mammograms were compared. In the third study, a secondary analysis was undertaken to assess the influence that prior mammograms have on the detection of different breast lesion types. Performance analyses for each lesion type were conducted separately for radiologists/breast physicians and radiographers. Statistical significance was defined as p ≤ 0.05.

Results: Prior mammograms improved radiologists’ specificity in dense and non-dense breasts (p≤0.01) and reduced false positives (p=0.01) but had no effect on sensitivity (p=0.37), lesion sensitivity (p=0.67), ROC (p=0.16), and JAFROC (p=0.24). Access to prior mammograms also improved radiographers’ specificity (p=0.002), ROC (p=0.003), and JAFROC (p=0.01) and reduced false positives (p=0.002) without affecting sensitivity (p=0.70) and lesion sensitivity (p=0.82). Both radiologists and radiographers’ performances were not influenced by breast density or their characteristics. Prior mammograms marginally improved radiologists’ detection of stellate/spiculated lesions.

Conclusions: Access to prior mammograms improves the ability to discriminate between normal and abnormal mammograms and reduces false positive decisions without affecting the detection of breast cancer, regardless of the characteristics of the mammogram reader and breast density.