Tomography for Intraoperative Evaluation of Breast Tumor Margins: A Simulation Study Qingsong Yang1, Ge Wang1, Gary Ge2, Xiaoqin Wang2, and Jie Zhang2* Department of Biomedical Engineering, Rensselaer Polytechnic Institute, Troy, NY, USA Department of Radiology, College of Medicine, University of Kentucky, Lexington, USA Introduction Breast conserving surgery involves an excision to achieve clear margins. This goal is usually achieved by performing intraoperative margin assessment with specimen imaging and immediate excision of any margins thought to be positive. However, routine two-dimensional specimen radiography cannot clearly discern tumor residual at the margin. Quasi-three-dimensional digital breast tomosynthesis potentially increases sensitivity but still cannot sufficiently meet the need, due to its cross-sectional limitation and strong image artifacts. Here we propose a novel tomographic technique that uses limited-angle tomosynthesis projections orthogonally acquired to reconstruct 3D transverse images. angles. These projections can then be reconstructed into sections as thin as 1 mm in the planes parallel to the detector. DBT allows 3D estimation of a tissue feature distribution, reduces the ambiguity from overlapping breast tissues, and improves details of noncalcified mammographic findings. Figure 3. Paired tomosynthetic scans for dual-limited-angle image reconstruction. The scanning scheme is on the top; the original 2D image created to mimic a breast tumor is on the lower left; and the reconstructed image on the bottom right. Figure 2. Schematic diagram showing the imaging principle of a digital breast tomosynthesis system. Figure 1. Exemplary intraoperative specimen radiographic images with positive margins (indicated by arrows) from a 61 year-old breast cancer patient. The left image shows pre-surgical needle localization; the middle and the right images show two orthogonal views of the specimen. Due to limitations of the 2D images, the patient underwent a second surgery for positive margins 20 days later. Simulation Result A 2D numerical phantom was created to mimic a breast tumor, as shown in Figure 3. This phantom was used to evaluate the proposed system. The phantom was defined on a 350x350 grid over a region of 5x5cm. Thus, the pixel size of the image was 143 µm. The geometric parameters were set in reference to a standard clinical DBT protocol. In total, 15 projections were acquired from -7.5˚ to 7.5˚. A SART-TV algorithm was used to reconstruct images. Figure 4. Three-view limited-angle reconstruction. The scanning scheme is on the top; the original 2D image created to mimic abreast tumor is on the lower left; and the reconstruction result on the bottom right. References R. E. Alvarez, etc., “Energy-Selective Reconstructions in X-Ray Computerized Tomography,” Phys Med Biol 21, 733-744 (1976). L. Yu, etc., “Dual-energy CT-based monochromatic imaging,” AJR Am J Roentgenol 199, S9-S15 (2012). T. Niu, etc., “Iterative image-domain decomposition for dual-energy CT,” Med Phys 41, 041901 (2014). T. P. Szczykutowicz, etc., “Dual energy CT using slow kVp switching acquisition and prior image constrained compressed sensing,” Phys Med Biol 55, 6411-6429 (2010). C. Maass, etc., “Image-based dual energy CT using optimized precorrection functions: a practical new approach of material decomposition in image domain,” Med Phys 36, 3818-29 (2009). Methodology Digital breast tomosynthesis (DBT) is an emerging technology that generates quasi-3D images by performing a series of low-dose radiographic exposures. Specifically, the x-ray tube moves in an arc overhead, generating a series of projections at preset