"3D Reconstruction of Long Object not Fitted in the Field of View of CT-Scanner using Sparse View Helical Scan Geometry"

"Long-object reconstruction under sparse-view helical computed tomography (CT) remains challenging due to truncation artifacts, noise degradation, and contrast loss. This manuscript presents a reconstruction frame work that addresses the instrumentation and measurement challenges associated with long-object imaging in helical CT. The framework emphasizes sparse-view acquisition, and this emphasis reduces scanning time and radiation exposure while preserving measurement efficiency. The study employs the Single-View Multiplicative Algebraic Reconstruction Technique (SvMART) as the core algebraic solver. It adapts for long-object scenarios through an proposed upscaling and downscaling strategy using Long object Reconstruction using Scale Adaption (LORSA). The upscaling operation stabilizes voxel\u2013ray interactions under truncation, while the downscaling operation restores the reconstructed volume to the original resolution. The recon struction process maintains numerical stability under lim ited projections. It supports long-object imaging beyond the scanner FOV. The framework further incorporates a post reconstruction enhancement module using the proposed Long object Reconstruction using Single Iteration (LORSI). It suppresses noise and improves structural clarity without modifying reconstruction geometry. The combined strategy produces improved reconstruction quality using fewer pro jections, thereby supporting efficient long-object imaging under constrained measurement conditions. Quantitative evaluation using the Structural Similarity Index, Jaccard Similarity Coefficient, Hausdorff Distance, and Chamfer Distance are performed. These show consistent improve ments from SvMART TO LORSA i.e., 5.8%, 9%, 8.6%, 7.36%, respectively and SvMART to LORSI are 28.7%, 24.3%, 21.94% and 29.91%, respectively. It increases structural fidelity and boundary accuracy. The proposed framework enables accurate reconstruction using fewer projection views, thereby reducing radiation dose, improv ing scanner efficiency, and extending system lifespan."

稀疏视图螺旋计算机断层扫描（CT）下的长物体重建仍面临截断伪影、噪声退化与对比度损失等挑战。本文提出一种重建框架，可解决螺旋CT长物体成像中涉及的仪器与测量难题。该框架以稀疏视图采集为核心，在保障测量效率的同时，可缩短扫描时长并降低辐射暴露剂量。本研究采用单视图乘性代数重建技术（Single-View Multiplicative Algebraic Reconstruction Technique，SvMART）作为核心代数求解器，并通过所提出的基于尺度自适应长物体重建（Long object Reconstruction using Scale Adaption，LORSA）的尺度升降策略，适配长物体成像场景。升尺度操作可在截断伪影影响下稳定体素与射线的交互作用，而降尺度操作则可将重建体积恢复至原始分辨率。该重建过程在有限投影数下可保持数值稳定性，且支持超出扫描仪视场（Field of View，FOV）的长物体成像。该框架还集成了基于所提出的单迭代长物体重建（Long object Reconstruction using Single Iteration，LORSI）的重建后增强模块，可在不改变重建几何参数的前提下抑制噪声并提升结构清晰度。该联合策略可在更少投影数下获得更高质量的重建结果，从而在受限测量条件下实现高效长物体成像。本研究采用结构相似性指数（Structural Similarity Index）、雅卡尔相似系数（Jaccard Similarity Coefficient）、豪斯多夫距离（Hausdorff Distance）以及倒角距离（Chamfer Distance）开展定量评估。评估结果显示，从SvMART到LORSA，各项指标分别提升5.8%、9%、8.6%与7.36%；从SvMART到LORSI，各项指标分别提升28.7%、24.3%、21.94%与29.91%。该方法可提升结构保真度与边界精度。所提出的重建框架可在更少投影视图下实现精准重建，从而降低辐射剂量、提升扫描仪运行效率并延长系统使用寿命。