Initial Clinical Experience with Stationary Digital Breast Tomosynthesis

Abstract

Rationale and objectives: A linear array of carbon nanotube-enabled x-ray sources allows for stationary digital breast tomosynthesis (sDBT), during which projection views are collected without the need to move the x-ray tube. This work presents our initial clinical experience with a first-generation sDBT device.

Materials and methods: Following informed consent, women with a "suspicious abnormality" (Breast Imaging Reporting and Data System 4), discovered by digital mammography and awaiting biopsy, were also imaged by the first generation sDBT. Four radiologists participated in this paired-image study, completing questionnaires while interpreting the mammograms and sDBT image stacks. Areas under the receiver operating characteristic curve were used to measure reader performance (likelihood of correctly identifying malignancy based on pathology as ground truth), while a multivariate analysis assessed preference, as readers compared one modality to the next when interpreting diagnostically important image features.

Results: Findings from 43 women were available for analysis, in whom 12 cases of malignancy were identified by pathology. The mean areas under the receiver operating characteristic curve was significantly higher (p < 0.05) for sDBT than mammography for all breast density categories and breast thicknesses. Additionally, readers preferred sDBT over mammography when evaluating mass margins and shape, architectural distortion, and asymmetry, but preferred mammography when characterizing microcalcifications.

Conclusion: Readers preferred sDBT over mammography when interpreting soft-tissue breast features and were diagnostically more accurate using images generated by sDBT in a Breast Imaging Reporting and Data System 4 population. However, the findings also demonstrated the need to improve microcalcification conspicuity, which is guiding both technological and image-processing design changes in future sDBT devices.

Keywords: 3D mammography; Breast cancer imaging; Digital mammography; Stationary digital breast tomosynthesis.

Figure 1:

Stationary digital breast tomosynthesis (sDBT). (A) Schematic representation of sDBT. Conventional DBT systems move a single standard x-ray tube through space to collect the projection views for reconstruction. In contrast, sDBT uses a distributed array of fixed carbon nanotube (CNT)-enabled x-ray sources to collect the projection views without any source motion. (B) The sDBT device in the University of North Carolina Cancer Hospital.

Figure 2:

Entrance dose as a function of breast thickness for the Senographe Essential mammography system (grey diamonds) and the stationary digital breast tomosynthesis (sDBT) system (black line). Doses are reported as the incident air kerma (mGy) and were measured using a mammography ion chamber dosimeter for sDBT. For each mammogram, doses were obtained from the DICOM header.

Figure 3:

Overall reader performance. The receiver operating characteristic (ROC) curves for Readers 1-4 when interpreting mammograms (A) and stationary digital breast tomosynthesis (sDBT) images (B). Sensitivity and specificity were calculated from the reported likelihood of malignancy at the time of interpretation and the actual presence of malignancy as determined by biopsy. The area under the ROC curve (AUC) provided a measure of diagnostic accuracy, and the mean AUC for sDBT was significantly higher than the mean AUC for mammography (C). AUCs were also calculated for each reader and each modality using a mixed-effect model (D). * represents p<0.0001.

Figure 4:

Average reader performance as a function of breast density (A) and compressed breast thickness (B). The area under the receiver operator characteristic curve (AUC) provided a measure of diagnostic accuracy. Readers were more likely to make a correct diagnosis using stationary digital breast tomosynthesis (sDBT) compared to mammography for each breast density category (BIRADS A-D) and breast thickness range, as reflected by statistically higher (p<0.05) mean AUCs.

Figure 5:

Reader confidence. (A) Average reader confidence in the overall impression when interpreting mammograms (gray) and stationary digital breast tomosynthesis (sDBT) images (black). (B) Readers were significantly more confident in their interpretation of images containing malignant lesions (darker shade) compared to their interpretation of images with benign lesions (lighter shade), with malignant and benign determined by pathology (* represents p<0.05, with bars representing the standard error of each group). This finding was similar for both mammography and sDBT.

Figure 6:

Reader preference. Bar graphs summarizing aggregate reader preference when interpreting diagnostically-important image features as displayed in the mammogram and stationary digital breast tomosynthesis (sDBT) image stack. Readers preferred sDBT over the mammogram when interpreting soft-tissue features (mass margins and shape, architectural distortion, and asymmetry) but preferred mammography when characterizing microcalcifications (p<0.05 in all cases, with error bars representing the 95% confidence interval of each grand mean estimate). Positive scores represent a preference for sDBT, and negative scores represent a preference for mammography.

Figure 7:

Example mammogram (A) and image slice from the reconstructed stationary digital breast tomosynthesis (sDBT) image stack (B). Based on pathology, the site of concern (expanded view) was benign. Readers were more likely to characterize this lesion accurately and were more confident in their assessment when interpreting sDBT, given the fact that the margins of the mass were more difficult to characterize on the mammogram. For this example, readers scored an average likelihood of malignancy of 50% when interpreting the mammogram and 25% when interpreting the sDBT images. The difference in microcalcification image intensity between the sDBT image slice and mammogram reflects the differences in the processing and display of each image. Window/level settings were selected to optimize feature display in the image overall.