Micro-computed tomography (µCT) is a non-destructive imaging tool that is instrumental in assessing bone structure in preclinical studies, however there is a lack of consensus on µCT procedures for analyzing the bone healing callus. This study provides a step-by-step µCT protocol that allows the monitoring of fracture healing.
Micro-computed tomography (µCT) is the most common imaging modality to characterize the three-dimensional (3D) morphology of bone and newly formed bone during fracture healing in translational science investigations. Studies of long bone fracture healing in rodents typically involve secondary healing and the formation of a mineralized callus. The shape of the callus formed and the density of the newly formed bone may vary substantially between timepoints and treatments. Whereas standard methodologies for quantifying parameters of intact cortical and trabecular bone are widely used and embedded in commercially available software, there is a lack of consensus on procedures for analyzing the healing callus. The purpose of this work is to describe a standardized protocol that quantitates bone volume fraction and callus mineral density in the healing callus. The protocol describes different parameters that should be considered during imaging and analysis, including sample alignment during imaging, the size of the volume of interest, and the number of slices that are contoured to define the callus.
Micro-computed tomography (µCT) imaging has been widely used in preclinical bone research, providing noninvasive, high-resolution images to evaluate the microstructure of bones1,2,3,4,5. µCT involves a large number of X-ray images, obtained from a rotating sample or by using a rotating X-ray source and detector. Algorithms are used to reconstruct 3D volumetric data in the form of a stack of image slices. Clinical CT is the gold standard for 3D imaging of human bones, and µCT is a commonly used technique for evaluating bone healing efficiency in experimental animals1,2,3,4,6,7. Mineralized bone has excellent contrast to X-ray, while soft tissues have relatively poor contrast unless a contrast agent is used. In the assessment of fracture healing, µCT generates images that provide detailed information about the 3D structure and density of the mineralized callus. In vivo µCT scanning can also be used for longitudinal, time-course assessment of fracture healing.
The quantification of intact cortical and trabecular bone using µCT is generally well-established and standardized8. Although preclinical studies use a variety of quantification methodologies to analyze fracture healing9,10,11, a detailed protocol of µCT image analysis for callus quantification has not been published yet.Therefore, the aim of this study is to provide a detailed step-by-step protocol for µCT imaging and analysis of bone healing callus.
The following protocol was developed to characterize long-bone healing callus harvested from euthanized mice. However, most of the steps can be applied to rats and also used for in vivo scanning of fractured bones. The protocol describes a particular µCT system and specific image processing, analysis, and visualization software (see Table of Materials), yet the methodology is generally applicable to other scanners and software. The protocol was approved by the Institutional Animal Care and Use Committee of The Pennsylvania State University College of Medicine. Mice used in this study were 16-week-old, male C57BL/6J mice (average weight 31.45 ± 3.2 g).
1. Tissue harvesting and preservation
NOTE: Use a suitable murine fracture model. For this study, the mid-diaphyseal open tibial fracture model was used according to the standard protocol described in12,13.
2. µCT scanning
Figure 1: Structure of the customized scanning fixture. (A) Images of the scanning fixture (top), showing the six sample slots, and the HA phantom (bottom). (B) Images showing the long-bone sample (top) and the HA phantom (bottom) placed in the dedicated slots. (C) Images showing the scanning fixture placed in a 20 mm syringe. Please click here to view a larger version of this figure.
3. Image segmentation
NOTE: Raw images are automatically reconstructed to image sequence data.
Figure 2: Image segmentation. (A) An image showing six samples within one scan. (B) Image cropping to isolate individual samples. (C) Digital alignment to correct a misaligned longitudinal axis (yellow dotted line). (D) Definition of the VOI and callus center plane. Please click here to view a larger version of this figure.
4. Image analysis
Figure 3: Segmentation of the callus outer boundary. (A) A contour of the outer boundary of the callus (red line). (B) Contours at slices sampled across the VOI (red slices). (C) A 3D callus label created by interpolation (red volume). (D) A cross-section of the callus label shown in C (including cortical bone). Please click here to view a larger version of this figure.
Figure 4: Segmentation of the cortical bone. (A) A contour of the periosteal surface of the cortex (green line). (B) Contours at slices sampled across the VOI (green slices). (C) A 3D label of the cortical bone (containing the medullary cavity; green) and the callus (red) created from interpolated labels of the periosteal cortex and the callus. (D) A cross-section of the callus (red) and the cortical bone (containing the intramedullary cavity; green). Please click here to view a larger version of this figure.
Figure 5: Conversion of gray-scale units to BMD. (A) Contours of the HA cylinder at the first and the last slices (red circles). (B) 3D interpolated HA cylinders (left) and cross-sections (right). Brown: highest HA density; blue: second highest HA density; violet: third highest HA density; green: fourth highest HA density. Please click here to view a larger version of this figure.
Figure 6: Segmentation of the mineralized callus. (A) The mineralized callus (≥250 mgHA/ccm) is shown in blue, the rest of the callus (<250 mgHA/ccm) is shown in red, and the space corresponding to the original bone is shown in green. (B) A 3D view of each isolated label. Please click here to view a larger version of this figure.
To monitor bone formation during fracture healing, a mid-diaphyseal open tibial fracture was induced in adult, male C75BL/6J mice. The fracture was stabilized using an intramedullary nail, an established model of secondary healing13. Callus tissues were harvested at days 14, 21, and 28 post-fracture12. These timepoints represent different phases of healing. Endochondral bone formation during secondary bone healing proceeds via initial formation of a fibro-cartilaginous (soft) callus, which mineralizes at later stages to reduce micromotion at the fracture gap, allowing the formation of new blood vessels across the fracture line13. Day 14 post-fracture in the murine fracture model used in this study represents the stage of mineralized soft callus. As healing proceeds from day 14 to day 21, the mineralized soft callus is completely replaced by newly formed woven bone, resulting in bony bridging of the fracture gap13. Between days 21 and 28, the callus undergoes resorption and remodeling to re-establish the characteristic structure of cortical bone12.
µCT images were acquired and analyzed at three timepoints using the protocol described above. A minimum of 10 samples were analyzed at each timepoint. For each sample, bone volume fraction and BMD were calculated. Bone volume fraction was calculated by dividing the volume of mineralized callus (BV) by the total callus volume (TV). Results demonstrated substantial formation of mineralized callus at day 14 (Figure 7A,B) and incremental increases in bone fraction volume and BMD as healing proceeded from day 14 to days 21 and 28 (Figure 7A,B), consistent with bony bridging of the fracture gap. As expected, the callus underwent resorption/remodeling between days 21 and 28 as evidenced by a decline in total callus volume (Figure 7A,B). Cortical bridging of the callus was more evident at day 28 than any preceding timepoint (Figure 7A). These results indicate that the provided µCT protocol allows monitoring of bone formation and callus structure during different phases of bone healing.
Figure 7: Monitoring bone healing using µCT. (A) 2D (sagittal, left panel) and 3D (right panel) images of the healing callus generated by µCT at the indicated post-fracture timepoints. (B) BMD, bone volume fraction (BV/TV), and total callus volume calculated from images shown in A. Results show healing progression through the late repair and remodeling phases. N = 10-12. The dots on the line plot represent average ± SEM. (*) p < 0.05 using one-way ANOVA followed by Tukey's post-hoc test. Please click here to view a larger version of this figure.
The purpose of this study is to describe a detailed protocol for µCT analysis with the goal of accurate quantification of the 3D mineralized callus structure, which is often fundamental in bone and fracture healing studies. The protocol utilizes a general-purpose state-of-the-art 3D image analysis software platform which facilitates image visualization, segmentation/labelling, and measurements ranging from simple to complex.
The most time-consuming task in the protocol is semi-automated segmentation of the callus, with exclusion of the cortical bone and medullary canal. This region has also been excluded in many previous studies9,16,17,18. Some studies have included the native cortical bone and canal regions in their analyses19,21, while in other studies the approach was not clear. Including the native cortices avoids the difficulty and potential subjectivity in contouring comminuted regions of fractured cortices but inflates callus mineralization measures.
The protocol focuses on obtaining output measures including total callus volume, mineralized volume, bone volume fraction, and bone mineral density. These parameters are readily interpreted and are commonly reported in the literature. Mineralized volume and bone volume fraction are dependent on the selected threshold for differentiating mineralized versus unmineralized, whereas bone mineral density is not. Tissue mineral density may also be computed based on only the tissue labeled as mineralized, instead of bone mineral density based on both mineralized and unmineralized callus. Tissue mineral density has been reported to be associated with torsional strength and rigidity9; however, these measures are more likely affected by partial volume effects and imaging resolution than bone mineral density.
Investigators have reported good correlation between quantified 3D cortical bridging and callus strength and stiffness (cortical bridging assessed on 2D radiographs is commonly assessed clinically in human patients)20. Additional 3D callus properties reported in the literature include moments of inertia10,15,19, which characterize the geometric distribution of the callus (i.e., how spread out the tissue is). Polar moment of inertia theoretically relates to torsional resistance and bending moment of inertia relates to bending resistance. Although these properties could be computed based on the segmented callus data described in this study, their correlation with measured biomechanical properties has been reported to be inconsistent9,19,21. Other previously reported callus properties include connectivity density, trabecular thickness, and structure model index11,17,,22. These parameters are often used to characterize trabecular bone and are readily computed by µCT scanner software; however, their relationship to fracture healing quality is not as clear. The software utilized in this protocol is a general-purpose program, not specific to bone. Thus, if certain bone parameters such as trabecular thickness are calculated outside of this protocol, segmented data may be exported to other programs for further analysis (e.g., as in Watson et al.23).
This protocol provides detailed workflows for complex callus structure characterization and quality control from a single software environment as compared to other methods in which multiple programs are required for analysis24. Therefore, time saving is a potential advantage of this protocol. The software enables a variety of flexible, sophisticated 3D visualization methods that help ensure accurate analysis and also allow for parallel tabulation of all results.
The µCT analysis protocol can be adapted to different fracture models in mice as well as rats; for other applications, optimization of some of the critical steps is recommended to ensure minimizing result variation. Specifically, investigating the impact of changing the size of the VOI or the number of contoured slices within the VOI on the reproducibility of the results should be considered. Also, using digital realignment as described in step 3.4 is recommended, but if different software is used for analysis, then assessing the necessity of this step, by comparing data generated with and without digital realignment, might be required.
In this protocol, a semi-automated segmentation approach was used for the identification and separation of the callus from the cortical bone and marrow. In cases like comminuted fractures, where the structure of the callus is extremely complex, contouring the callus and the periosteal surface of the cortex becomes challenging. It is advisable in these cases to perform the contouring with multiple experimenters to assess and attempt to limit subjectivity.
Limitations exist with this protocol. The protocol requires the conversion and export of DICOM images so that images can be subsequently analyzed in additional software; this step takes some additional time and may necessitate use of a calibration phantom within the image. As automated segmentation techniques continue to evolve, including ones based on machine learning, it may be advantageous to replace the manual contouring portions of the protocol with these new techniques. Overall, the detailed protocol described here for the analysis of bone healing callus in rodents may especially benefit labs without substantial µCT analysis experience and may help establish a more consistent and standardized approach across the field.
The authors have nothing to disclose.
This work was supported by National Institutes of Health (NIH) R01 DK121327 to R.A.E and R01 AR071968 to F.K.
10% neutral buffered formalin | Fisher chemical | SF100-20 | Used for bone tissue fixation |
Avizo | Thermo Scientific | Image processing and analysis software | |
Hydroxyapatite phantom | Micro-CT HA D4.5, QRM | QRM-70128 | |
Image Processing Language | Scanco | Used to convert raw images to DICOM images | |
Micro-Mosquito Straight Hemostatic Forceps | Medline | Used to remove the intramedullary pin | |
Microsoft Excel | Microsoft | Spreadsheet software | |
Scanco mCT system (vivaCT 40) | Scanco | Used for µCT imaging |