Purpose: To investigate how the greater contrast of full-field digital mammography (FFDM) affects the detection of suspicious lesions in Japanese population-based screening.
Materials and methods: Screen-film mammography (SFM) and FFDM were performed in 480 women aged 50 years or more. A set of mediolateral oblique views was obtained with each modality. All mammograms were independently double-read. The five-scale category assessment and type of finding using the Breast Imaging Reporting and Data system (BI-RADS) nomenclature were given. Intraobserver variance, recall rates, and positive predictive value were calculated.
Results: The findings between the two modalities were discordant. kappa-values for each reader were 0.619 and 0.385, respectively. Almost half of the microcalcifications were called with both modalities. The detection of masses was less concordant between the readers (27%). The masses were detected more frequently with FFDM (73%). Other findings were only detected with one modality. The recall rate was not significantly different (2.9% with SFM vs. 4.2% with FFDM; p=0.253). The positive predictive value was not significantly different (14% with SFM vs. 10% with FFDM; p=0.69), either. Two patients with breast cancer were detected with both modalities.
Conclusion: Recall rates and positive predictive value were not significantly different between SFM and FFDM. Cancers were detected with both modalities.