An ideal imaging technique for lung nodule screening would allow the visualization of small nodules within a complex anatomical background, use a low radiation dose technique, acquire images in <0.25s, and retain patient positioning flexibility. A novel C-arm mounted scanning beam x-ray source and digital detector system (SBDX) can acquire tomosynthesis images in real time, We investigate, using numerical simulation, this approach for lung nodule detection. A high-resolution CT volume (0.5 mm isotropic voxels) of a plastinated dog lung was acquired. Spherical nodules (40 HU) and overlying ribs (cortical bone1000 HU) were added numerically, providing a detection task with typical anatomic complexity. Tomographic blurring was modeled by convolving each slice with a normalized cylindrical blur function (edges rolled off using cosines). Lesion visibility was examined as a function of tome-angle and lesion size. For lesions 4.5 mm and 2.5 mm in diameter, half-tomo-angles of at least 3<degrees> and 4.5 degrees respectively are necessary before visualization of the lesions improves. Modification of the SBDX system (current half-tome-angle = 1.5 degrees) is therefore desired before optimal lung nodule detection is feasible. Possible approaches include increasing the size of the digital detector, and decreasing the object-to-detector distance.