Evaluation of the image quality in digital breast tomosynthesis (DBT) employed with a compressed-sensing (CS)-based reconstruction algorithm by using the mammographic accreditation phantom

In this work, we have developed a prototype digital breast tomosynthesis (DBT) system which mainly consists of an x-ray generator (28 kVp, 7 mA s), a CMOS-type flat-panel detector (70-μm pixel size, 230.5×339 mm2 active area), and a rotational arm to move the x-ray generator in an arc. We employed a compressed-sensing (CS)-based reconstruction algorithm, rather than a common filtered-backprojection (FBP) one, for more accurate DBT reconstruction. Here the CS is a state-of-the-art mathematical theory for solving the inverse problems, which exploits the sparsity of the image with substantially high accuracy. We evaluated the reconstruction quality in terms of the detectability, the contrast-to-noise ratio (CNR), and the slice-sensitive profile (SSP) by using the mammographic accreditation phantom (Model 015, CIRS Inc.) and compared it to the FBP-based quality. The CS-based algorithm yielded much better image quality, preserving superior image homogeneity, edge sharpening, and cross-plane resolution, compared to the FBP-based one.