Comparative validation of reading 1-mm-thick and 6-mm-thick images in breast tomosynthesis

At our institution, breast cancer screening with digital breast tomosynthesis (DBT) has been implemented since 2017. The addition of DBT was found to improve the process index. However, the increase in exposure dose and imaging time, as well as the increase in reading time and data volume were also identified as major problems. In this study, we compared the reading results of the 6-mm-thick images yielded by Hologic's new technology (3DQuarummTM) with those of conventional 1-mm-thick images.[Subjects and Methods] We used imaging data obtained from 443 outpatients who underwent digital mammography (2D) plus DBT (3D), and the lesions in 474 breasts were classified as category 3or higher. Two sets of images were prepared for each case, "2D + 1-mm-thick images 3 D" and "2D + 6-mm-thick images 3D," and three doctors interpreted the findings in both sets at intervals of at least one month.[Result] The category agreement rate for all cases was 77.5%, and the judgment agreement rate (for differentiating between category 2 or lower and category 3 or higher) was 85.9%, with no significant differences between the 1-mm-thick and 6-mm-thick images. The category agreement rate for cancer cases was 75.0%, and the judgment agreement rate was 90.5%. Category discrepancies in cancer cases tended were represented by the classification of mass lesions into higher categories in the 6-mm-thick images and into lower categories in the 1-mm-thick images, while the distortion rate was higher in the 1-mm-thick images than in the 6-mm-thick images. In regard to the reading time evaluated in 20 cases, the reading time was 2.7 minutes lower for the 6-mm-thick images than for the 1-mm-thick images.[Conclusion] A comparison between 6-mm-thick and 1-mm-thick images revealed a favorable agreement rate for discriminating between category 2 or less and category 3 or more lesions. A reduced time for the 6-mm-thick images was evident. Therefore, we consider that obtaining 6-mm-thick images is acceptable for the purpose of screening.