Experimental phantom lesion detectability study using a digital breast tomosynthesis prototype system

Abstract

Purpose: To compare the sensitivity of conventional two-dimensional (2D) projection imaging with tomosynthesis with respect to the detectability of mammographic phantom lesions.

Materials and methods: Using a breast tomosynthesis prototype based on a commercial FFDM system (Siemens MAMMOMAT Novation), but modified for a wide angle tube motion and equipped with a fast read-out amorphous selenium detector, we acquired standard 2D images and tomosynthesis series of projection views. We used the Wisconsin mammographic random phantom, model RMI 152A. The anode filter combinations Mo/Mo and W/Rh at two different doses were used as typical radiographic techniques. Slice images through the phantom parallel to the detector were reconstructed with a distance of 1 mm employing a filtered back-projection algorithm. The image data sets were read by five radiologists and evaluated with respect to the detectability of the phantom details.

Results: For all studied radiographic techniques, the detection rate in the tomosynthesis mode was 100 %, i. e. 75 true positive findings out of 75 possible hits. In contrast, the conventional projection mode yielded a detection rate between 80 and 93 % (corresponding to 60 and 70 detected details) depending on the dose and X-ray spectrum.

Conclusion: Tomosynthesis has the potential to increase the sensitivity of digital mammography. Overlapping structures from out-of-plane tissue can be removed in the tomosynthesis reconstruction process, thereby enhancing the diagnostic accuracy.