Influence of image reconstruction kernel on computed tomography-based finite element analysis in the clinical opportunistic screening of osteoporosis—A preliminary result

Abstract

Purpose: This study aimed to evaluate the difference in vertebral mechanical properties estimated by finite element analysis (FEA) with different computed tomography (CT) reconstruction kernels and evaluate their accuracy in the screening and classification of osteoporosis. Methods: There were 31 patients enrolled retrospectively from the quantitative CT database of our hospital, uniformly covering the range from osteoporosis to normal. All subjects’ CT raw data were reconstructed both with a smooth standard convolution kernel (B40f) and a sharpening bone convolution kernel (B70f), and FEA was performed on L1 of each subject based on two reconstructed images to obtain vertebral estimated strength and stiffness. The trabecular volumetric bone mineral density (vBMD) of the same vertebral body was also measured. FEA measurements between two kernels and their accuracy for osteoporosis screening were compared. Results: The vertebral stiffness and strength measured in FEA-B40f were significantly lower compared with those of FEA-B70f (12.0%, p = 0.000 and 10.7%, p = 0.000, respectively). The correlation coefficient between FEA-B70F and vBMD was slightly higher than that of FEA-B40F in both vertebral strength and stiffness (strength: r2-B40f = 0.21, p = 0.009 vs. r2-B70f = 0.27, p = 0.003; stiffness: r2-B40f = 0.37, p = 0.002 vs. r2-B70f = 0.45, p=0.000). The receiver operator characteristic curve showed little difference in the classification of osteoporosis between FEA-B40f and FEA-B70f. Conclusion: Two kernels both seemed to be applicable to the opportunistic screening of osteoporosis by CT-FEA despite variance in FE-estimated bone strength and bone stiffness. A protocol for CT acquisition and FEA is still required to guarantee the reproducibility of clinical use.