Purpose: To assess the clinical translational potential of a phase sensitive breast tomosynthesis (PBT) system through a pilot observer preference and performance study.
Materials and Methods: The first-of-its-kind prototype PBT system was developed for patient imaging based on the inline phase contrast principle with a rotating gantry and a microfocus polyenergetic x-ray tube. Fifty-two patients recalled with suspicious breast lesions detected on screening digital breast tomosynthesis (DBT) were recruited for additional PBT imaging before lesion biopsy. PBT images were acquired at lower radiation doses than DBT. Nine radiologists independently evaluated image quality and classified lesions by retrospectively evaluating DBT and PBT images in two separate reading sessions. Image quality was subjectively rated. Areas under receiver operating characteristics curves (AUC) were used to assess and compare lesion classification performance of radiologists.
Results: For image appearance and general feature visibility rated by observers, DBT images have significantly higher scores on average than PBT images (p < 0.01). However, for lesion classification, the mean AUC value of reading PBT images is 0.781, which is slightly higher than the mean AUC value of 0.774 in reading DBT images (p = 0.84).
Conclusion: The new PBT modality reduces the radiation dose on average by 24% as compared to DBT. Although current PBT images have lower image quality, the slightly higher (non-significant) mean AUC values indicate that reading PBT images could potentially yield the same diagnostic accuracy as reading the standard of care DBT images. Additionally, the appearance of the PBT images can be improved by reducing the image scanning time and optimizing the image reconstruction and display software, both of which are feasible and planned for future studies.
Clinical Relevance Statement: Phase contrast breast imaging could potentially provide the same diagnostic accuracy as standard-of-care DBT at an average dose reduction of 24%.
