Basics of Digital Breast Tomosynthesis

Methods of Acquiring DBT Images

Similar to a standard DM acquisition, with DBT, the breast remains stationary in compression and is repositioned for each view. However unlike DM where the gantry is stationary during the exposure, DBT images are acquired while the gantry moves in an arc across the breast to obtain multiple very low-dose images at different angles. The number of projection images obtained per view are fixed (manufacturer determined) and are independent of breast thickness. The “reconstructed slices” are generated from the projection images and are the images that the radiologist interprets. The thickness of these images can be manipulated by the reader in some systems or is determined by the manufacturer. The number of reconstructed slices is dependent on the thickness of the breast (i.e., if the breast compresses to 5 cm and the reconstructed slices are 1-mm thick, then there will be 50 images to interpret for that view).

Two main methods of obtaining the projection images are the “step and shoot” method and the “continuous” method ( Box 6.1 ). As the names imply, with the step and shoot method, the gantry comes to a full stop between each low-dose exposure. With the continuous method, the gantry sweeps across the breast in one motion and acquires multiple images during the exposure. The continuous method has a faster acquisition (and therefore less patient motion), but gantry movement during exposure results in some focal spot blur. The main advantage of the step and shoot method is decreased focal spot blur, resulting in improved signal to noise. However, because the gantry is stopping for each exposure, there is increased time of acquisition and the potential for patient motion.

Imaging With DBT in One View Only

Several studies have shown noninferiority of acquiring DBT in one view and DM in the other view as a dose-reduction method. Because it has been shown that cancers were rated more visible on DBT compared with DM, imaging the breast in the mediolateral oblique (MLO) view with DBT increases the opportunity for cancer detection. However, despite the improved conspicuity of findings, imaging with DBT in both views provides the greatest opportunity for lesion detection. In fact, cancers have been shown to be more conspicuous on the DBT craniocaudal (CC) view compared with the DBT MLO view, which could lead to missed cancers if DBT is only performed on the MLO view. Therefore imaging with DBT for both views is optimal.

Synthetic Imaging

The FDA has approved software advancements that generate two-dimensional (2D) SM from the DBT data set. Because several studies have demonstrated that synthetic images are nearly as good as a standard DM, a number of facilities have stopped acquiring the DM image, effectively reducing the dose of the DM + DBT to approximate that of original DM levels.

Because DBT is the source of the SM image, there is often enhanced visualization of DBT-only findings such as architectural distortion and masses. Additionally, calcifications on SM are brighter than on DM and contribute to improved interpretation speed. Importantly, because the SM images are effectively a summation of the DBT reconstructed slices, the benefit of the thin slices are not gained and overlapping tissue may still obscure findings. SM should serve as a preview of the findings on DBT and must not be a replacement for reviewing the DBT slices.

There have been several improvements over time with the quality of synthetic images. The earliest versions had problems with reconstruction that generated features of the breast that were pixelated, such as the nipple. Additionally, resolution of the morphology of calcifications was poor. Furthermore, earlier versions of SM were challenged with an artifact generated from the processing algorithm that artificially enhanced structural elements in the breast that falsely appeared to represent calcifications, termed pseudocalcifications . However, the most recent high-resolution synthetic images appear almost indistinguishable from a standard DM and result in very few pseudocalcifications. Research has demonstrated that the performance of synthetic images when used in conjunction with DBT approximates that of the combination examination. Additionally, better visualization of skin lines and improved conspicuity and characterization lesion margins are an added benefit of the higher resolution SM. Decreased dropout (“halo” artifact) adjacent to large calcifications or other high attenuation objects such as surgical clips is minimized. Finally, reduced background noise and image blur contribute to the image quality and approximation to DM. However, the very large file sizes of these higher-resolution SM images require significant storage and bandwidth to transfer data efficiently. Fig. 6.2 shows a DM and corresponding standard resolution and high-resolution SM for a patient.