Accelerated beam-blocker-based scatter correction for clinical CBCT via sparsely-sampled angular interpolation

Photon scattering-induced artifacts degrade the quantitative and qualitative reliability of cone-beam computed tomography (CBCT) images, restricting their clinical diagnostic utility. Conventional beam-blocker-based scatter correction methods, despite enabling direct and cost-effective scatter measurement, necessitate additional scatter projections, resulting in elevated radiation dose and prolonged scan duration-two critical bottlenecks for clinical translation. To overcome these limitations, this study proposes a low-dose, efficient beam-blocker scatter correction approach integrating indirect scatter correction (IDSC) with angular interpolation, tailored for clinical CBCT scenarios. Capitalizing on the intrinsic low-frequency angular property of scatter, the proposed method acquires beam-block projections only at limited angles and retrieves missing scatter samples via linear interpolation, which is validated to possess high robustness in both head and pelvis phantoms. The optimal number of beam-block projections (15 for head, 86 for pelvis) is determined via data-driven adaptive selection based on scatter variance, anatomical complexity, and the clinical acceptability threshold of RMSE <5 HU for soft tissue, with no new models or experiments introduced. This selection is visually validated by figures2(a)-6(a), confirming that 15 and 86 projections achieve minimal CT number Hounsfield unit (HU) error and no visible artifacts, while fixed 60 projections or further reduction to 40 projections fail to meet clinical diagnostic standards. Open projections were corrected using the fused measured-interpolated scatter data prior to CBCT reconstruction. Phantom validation results demonstrated outstanding performance: for the head phantom, reducing beam-block projections from 600 to 15 achieved 97.5% dose/time savings with a post-correction CT number error (CTNE) of 4.13 ± 0.32 HU; for the pelvis phantom, decreasing beam-block projections from 600 to 86 yielded 85.67% dose/time reduction, with a residual CTNE of 8.67 ± 0.40 HU. A key limitation is that the assumption of smooth angular scatter variation may be invalid in metallic implant environments, requiring combination with metal artifact reduction techniques. Collectively, the proposed integrated framework-combining an optimized beam blocker, sparse projection selection, and IDSC with angular interpolation-achieves an optimal balance between dose reduction, acquisition efficiency, and image fidelity, substantially mitigating the scatter measurement burden without compromising diagnostic accuracy. It offers a practical, translatable solution for optimizing clinical CBCT workflows.