Since the invention of modern CT systems, metal artifacts have been a persistent problem. Due to increased scattering, amplified noise, and insufficient data collection, it is more difficult to suppress metal artifacts in cone-beam CT, limiting its use in human- and robot-assisted spine surgeries where metallic guidewires and screws are commonly used. In this paper, we demonstrate that conventional image-domain segmentation-based MAR methods are unable to eliminate metal artifacts for intraoperative CBCT images with guidewires. To solve this problem, we present a fine-grained projection-domain segmentation-based MAR method termed PDS-MAR, in which metal traces are augmented and segmented in the projection domain before being inpainted using triangular interpolation. In addition, a metal reconstruction phase is proposed to restore metal areas in the image domain. The digital phantom study and real CBCT data study demonstrate that the proposed algorithm achieves significantly better artifact suppression than other comparing methods and has the potential to advance the use of intraoperative CBCT imaging in clinical spine surgeries.