The spatial resolution criteria for assessment of trabecular microstructure

Objectives The spatial resolution is one of the main parameters that dictates the quality of microCT datasets. Since high-resolution scanning requires long acquisitions times, as well as huge data processing, optimization strategies for achieving ideally small voxel size and analysis time is crucial. In other areas of research, a minimum ratio of object size to voxel size of 2 has been suggested. Here we investigated the minimum necessary resolution that still generates reliable data for different microstructural features of trabecular bone in a rat animal model.

Methods The 3rd lumbar vertebral body of female Sprague-Dawley rats (Harlan Winkelmann, Borchen, Germany; 25-week-old; n = 3) were scanned at 90 kVp source voltage and 88 µA, using a micro-CT system (µCT 50, SCANCO Medical AG, Switzerland). The trabecular thickness of rats is typically reported at 60 µm. Hence, we chose five different voxel sizes: 4.4, 7.4, 10.0, 14.8, and 34.4 µm resulting in object size to voxel size ratios between 13 and 2.43. The volume of interest was 7.5 mm3 in the mid-section of the vertebral body, which excluded the cortical bone. Bone morphometric parameters such as bone volume fraction (BV/TV), trabecular thickness (Tb.Th.), trabecular number (Tb.N.), connectivity density (Conn.D.), and bone mineral density (BMD) were then measured directly in 3D models. The mean values at 4.4 µm were considered as reference points. The minimum necessary resolution was defined as the lowest scan resolution (i.e. highest voxel size) where the differences of the morphometric parameters were not statistically different compared to the data at reference point. Statistical analysis was carried out using SPSS package (IBM SPSS, version 25.0). The level of significance was designated as follows: no significant difference p > 0.05 (NS), p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***)

Results and Conclusion We found evident trends for all morphometric parameters when the voxel size increased. That is, BV/TV, Tb.N., and Conn.D. reduced, Tb.Th. increased, while BMD showed inconsistent fluctuations. Up to 14.8 µm, the differences compared to the reference point (4.4 µm) were insignificant for all parameters, except for Conn.D., which already showed a significant decline at 14.8 µm (*). At 34.4 µm, however, BV/TV, Tb.N., Conn.D., and BMD were significantly 74.0 % (**), 64.7 % (***), 95.3 % (***), and 5 % (*) reduced, respectively. On the other hand, the value of Tb.Th. increased by 39.2 % (*). At 14.8 and 34.4 µm, the ratio of “Tb.Th./voxel size” were 4.0 and 2.4, respectively.

Our results indicate that massive changes of the data occur, if the resolution is too low. The minimal necessary resolution to examine the trabecular structure of rat bones is at least 10.0 µm for the Conn.D., and 14.8 µm for other main morphometric parameters. Hence, we conclude that the previously-suggested object to voxel size of 2 is not conservative enough, and propose a minimum “Tb.Th./voxel size” of 4 for accurate measurement of the trabecular microstructure.

Stichwörter micro computed tomography, resolution, microstructure, trabecular bone

Publication History

Article published online:
15 October 2020