Reconstruction-kernel Dependence of Femoral Bone Marrow Attenuation in Whole-body Dual-source Photon-counting CT of Patients with Multiple Myeloma.

RATIONALE AND OBJECTIVES: Reconstruction kernels have various effects on image characteristics. This study aims to investigate how kernel selection influences the femoral bone marrow attenuation in patients with multiple myeloma. MATERIALS AND METHODS: This retrospective study included 103 consecutive patients who underwent noncontrast whole-body photon-counting computed tomography (PCCT) for assessment of multiple myeloma. All examinations were performed in ultrahigh-resolution dual-source mode (96 × 0.2 mm) at 90 kV/Sn 150 kV. For each examination, standard body regular (Br) kernels from Br32 to Br68 and the corresponding quantitative kernels (i.e., Qr32-Qr68) were reconstructed from 70 keV virtual monoenergetic images. Regions of interest were placed bilaterally in the femoral bone marrow, the adjacent muscle, and subcutaneous fat. Marrow attenuation and image noise were compared across kernels. RESULTS: Compared to the quantitative kernel Qr40 as reference standard (median -66.9 [interquartile range -82.0--45.5] hounsfield units [HU]), marrow attenuation was substantially lower with Br32 (-95.9 [-113.1--73.4] HU), Br36 (-96.0 [-111.9--73.0] HU), and Br40 (-89.5 [-102.7--65.1] HU) (all p < 0.001). Meanwhile, attenuation values did not differ significantly among the sharper standard body kernels Br48 to Br68 (all p > 0.999). No attenuation difference was ascertained among the quantitative kernels, regardless of sharpness (all p ≥ 0.108). Muscle and fat attenuation remained stable across reconstructions (all p ≥ 0.077). Image noise increased progressively with higher kernel sharpness in both body and quantitative kernels (all p ≤ 0.007). CONCLUSION: Femoral bone marrow attenuation in dual-source PCCT is reconstruction-kernel dependent. Quantitative marrow assessment in patients with multiple myeloma, therefore, requires strict reconstruction standardization, and attenuation measurements should not be considered reconstruction-invariant. Otherwise, attenuation differences may reflect reconstruction choice rather than changes in marrow composition.