A Sectioning, Coring, and Image Processing Guide for High-Throughput Cortical Bone Sample Procurement and Analysis for Synchrotron Micro-CT

Podsumowanie

We employed a geological (coring) sampling protocol to procure cortical bone specimens of uniform size for SRµCT experiments from the anterior aspect of human femora. This method is minimally destructive, efficient, results in cylindrical specimens that minimize imaging artifacts from irregular sample shapes and improves microarchitectural visualization and analysis.

Streszczenie

Bone is a dynamic and mechanically active tissue that changes in structure over the human lifespan. The products of the bone remodeling process have been studied substantially using traditional two-dimensional techniques. Recent advancements in X-ray imaging technology via desktop micro-computed tomography (µCT) and synchrotron radiation micro-computed tomography (SRµCT) have allowed for the acquisition of high-resolution three-dimensional (3D) scans of a larger field of view (FOV) than other 3D imaging techniques (e.g., SEM) providing a more complete picture of microscopic structures within human cortical bone. The specimen should be accurately centered within the FOV, however, to limit the appearance of streak artifacts known to impact data analysis. Previous studies have reported procurement of irregularly shaped rectilinear bone blocks that result in imaging artifacts due to uneven edges or image truncation. We have applied a geological sampling protocol (coring) to procure consistently sized cortical bone core specimens for SRµCT experiments from the anterior aspect of human femora. This coring method is efficient and minimally destructive to tissue. It creates uniform cylindrical samples that decrease imaging artifacts by nature of being isometric during rotation and providing a uniform path length for X-ray beams throughout scanning. Image processing of X-ray tomographic data of cored and irregularly shaped samples confirms the potential of the technique to improve visualization and analysis of cortical bone microarchitecture. A goal of this protocol is to deliver a reliable and repeatable method for the extraction of cortical bone cores that is adaptable for various types of high-resolution bone imaging experiments. An overarching goal of the work is to create a standardized cortical bone procurement for SRµCT that is affordable, consistent, and straightforward. This procedure may further be adapted by researchers in related fields who commonly evaluate hard composite materials such as in biological anthropology, geosciences, or material sciences.

Wprowadzenie

With recent advancements in imaging technology, it is now feasible to acquire X-ray imaging data with very high resolution. Desktop micro-CT (µCT) systems are the current standard for imaging cancellous bone due to their non-destructive nature¹. When imaging microstructural features of cortical bone, however, µCT use has been more limited. Due to resolution constraints, desktop systems cannot attain the resolution required to image microstructural features smaller than cortical pores, such as osteocyte lacunae. For this application, SRµCT is ideal owing to the greater resolution of these systems¹. For example, experiments at the Canadian Light Source (CLS) on the BioMedical Imaging and Therapy (BMIT) beamlines² have produced images with voxels as small as 0.9 µm. Previous studies^1,3,4,5 have used this resolution to acquire projections and subsequent three-dimensional (3D) renders from cortical bone specimens from human long bones (Figure 1) to quantify osteocyte lacunar density^4,6,7,8,9 and variation in the lacunar shape and size³ across the human lifespan and between the sexes. Further studies have demonstrated the presence of osteon banding in humans¹⁰, a phenomenon previously recognized to be associated with only nonhuman mammals in the forensic anthropological literature.

In order to achieve exceptional resolution, the X-ray beam must be finely focused within the field of view (FOV), which often limits the maximum specimen size to a few millimeters in diameter. Currently, there have been no comprehensive, standardized procedures described in the literature outlining bone sample procurement that meet these restrictions. Centering specimens within the FOV is critical to ensure that 1) the sample remains centered as it rotates 180° during imaging, and 2) scan artifacts are limited since there is no image truncation. In other words, no portions of the sample outside of the FOV interfere with the beam entering its focal point inside the FOV. If this occurs, the reconstruction algorithm is deprived of some of the attenuation data needed for a fully correct reconstruction. It is further worth noting that 360° (full rotation) scans minimize the effects of beam hardening but increase artifacts caused by misalignment and sample movement during imaging. Thus, while a 360° scan will typically generate cleaner data, imaging time is doubled and so a compromise between experimental cost and data quality must be addressed.

An important and often overlooked aspect of bone imaging experiments is the accurate and replicable specimen preparation technique performed prior to scanning. Studies that incorporate SRµCT methods into their experiments briefly mention their sampling protocol, but the authors provide little to no detail regarding the particular methodology used to gather their specimens. Many such studies mention cutting rectilinear bone blocks of arbitrary dimensions, but generally provide no further information about the tools or embedding materials used^{3,4,10,11,12,13,14}. Some researchers commonly use handheld rotary tools (e.g., Dremel) to remove rectilinear blocks of bone from a region of interest (ROI)^{3,4,10,11,12,13,14}. This method results in nonuniformly sized samples that may be larger than the FOV, increasing the likelihood of scan artifacts and image truncation. Such specimens often require further refining using a precision diamond-wafer saw (e.g., Buehler Isomet). Procuring samples with consistent dimensions (to the two-hundredths/mm) is critical to ensure that the acquired datasets are of the highest quality and the subsequent results are replicable.

The limited reporting of sample procurement methodology adds an extra layer of difficulty when attempting to employ and/or validate methods performed in a previous study. Currently, researchers must contact authors directly for further details on their sampling procedures. The protocol detailed here provides biomedical researchers with a thoroughly documented, replicable, and cost-efficient sampling technique. The primary objective of this article is to provide a comprehensive tutorial regarding how to procure consistently sized cortical bone core samples using a mill-drill press and diamond coring bit for the accurate visualization and extraction of microarchitectural data. This method is modified from procedures used to routinely collect uniform, small-diameter (1-5 mm) cylinders from blocks of hard materials in high pressure rock mechanics^{15,16,17,18,19}.

Protokół

All specimens were sourced from embalmed cadaveric donors at the University of Toledo, College of Medicine and Life Sciences and Northeast Ohio Medical University (NEOMED), with the informed consent of the donor themselves or the donor's next-of-kin. The University of Akron Institutional Review Board for the Protection of Human Subjects (IRB) deemed these specimens exempt from full IRB review as they were not procured from living individuals. Demographic information including age, sex, and cause of death were available for all donors. The selected individuals did not have documented bone-affecting conditions nor exposure to treatment regimens that may have affected bone remodeling at the time of death. Cortical bone samples were obtained from femora of cadaveric modern males and females with ages ranging from 19 to 101 years of age (mean = 73.9 years). The femoral midshaft has been studied extensively including examinations of variation in cortical porosity^{20,21,22,23,24} and material density of bone tissue^25,26,27, and has thus become a commonly used site for microstructural analyses.

1. Tissue Procurement and Maceration

1. Use an oscillating saw equipped with a plunge cutting carbide blade (for composite materials) to procure ~7.5 cm bone blocks from the mid-diaphyses of the left femora.
2. Soak femoral blocks in an oven-safe glass dish filled with a powdered protease enzyme and tap water solution for 1 h in an incubator set at 45 °C.
3. After incubation, carefully remove any remaining soft tissues and periosteum using blunt dissection or dental tools.
   NOTE: Avoid the use of sharp tools (e.g., scalpel) to remove soft tissues. Such instruments may cause damage to the bone that is detectable in µCT scans, affecting specimen preservation and scan data quality.
4. Remove debris or occlusions in the medullary cavity by placing bone blocks in an ultrasonic cleaner for 5-10 minutes with 20:1 parts tap water to cleaning solution (see Table of Materials) or using a handheld water flosser (e.g., Waterpik).
5. Immerse the bone block in a specimen cup and fill with 70% ethanol. Allow the bone to soak for at least 24 hours to remove lipids.
   NOTE: Xylenes may also be used to remove lipids. Extended soaking in xylenes, however, may make the bone brittle or chalky as it is an emulsifier.
6. After 24 hours, remove bone blocks from ethanol and allow to air dry at ambient temperature for 24-48 hours.
   NOTE: The protocol may be paused here.

2. Tissue sectioning

1. Place a 75 x 25 mm glass microscope slide on a hot plate set to 140 °C. Melt a generous amount of thermal epoxy resin (see Table of Materials) on the center of the slide.
   1. If preparing additional thin sections for microscopy (<50 µm), the bone block may need to be embedded in a two-part epoxy to preserve trabeculae. Further, when implementing this protocol for fragile specimens (e.g., diagenetic bone or highly trabecularized specimens) embedding specimens in an epoxy is necessary.
      NOTE: Bone samples used in this protocol were retrieved from embalmed cadaveric specimens. If fresh specimens are collected at autopsy or from a surgical case to examine soft tissue structures (e.g., vasculature) via SRµCT, impregnation with epoxy may induce damage to such tissues. In these cases, an alternative adhesive or mounting medium is recommended (e.g., double-sided tape, modeling clay).
2. Press the inferior aspect of the bone block into the thermal epoxy resin on the microscope slide, with the length of the bone perpendicular to the slide. Shift the sample back-and-forth in order to coat the underside of the bone and ensure secure adhesion to the slide.
3. Let the mounted specimen rest on the hot plate for ~5 min to allow the thermal epoxy to wick into pores and/or cracks.
   NOTE: The epoxy on the slide should be free of bubbles for best adhesion. To remove bubbles, shift the sample back and forth on the slide. Bubbles often form due to water and/or ethanol trapped within the bone escaping and evaporating.
4. Remove the slide with the mounted specimen from the hot plate using blunt forceps and allow to cool at room temperature for ~10 minutes. Remove any epoxy from the edge of the slide using a razor blade to ensure the chuck adequately grips the slide.
5. Attach the slide with the adhered sample to a glass slide chuck and mount the chuck to the swivel arm of a slow-speed sectioning saw (see Table of Materials, Figure 2).
   NOTE: While a Buehler IsoMet saw was employed in this protocol, other precision sectioning saws are available that could be used in place of the IsoMet (e.g., Leco, Exakt, Smartcut, CT3, Buehler Petrothin, Well Diamond Wire).
6. Adjust the swivel arm using the positioning dial to ensure the blade contacts and transects the sample. Position the specimen such that a cross-section of the bone will be cut perpendicular to its length.
7. Add weights to the far side of the cutting arm to counter the weight of the arm.
   NOTE: If insufficient counterweight is used, the sample may bear down on the blade and cause the blade to fracture.
8. Add cutting fluid (20:1 parts water to cutting fluid) to the fluid receptacle of the saw.
9. Tightly secure the diamond wafer blade and ensure the fluid level submerges the cutting portion of the blade. Set the speed to 200 RPM and slowly lower the sample onto the blade (Figure 3).
10. Ensure the blade and chuck are not wobbling and/or bouncing. If excessive movement is noted, immediately stop the saw and tighten the blade and/or chuck arm assembly before resuming cutting. Add additional counterweights if the chuck is aggressively moving up and down. Excessive movement including visible side-to-side motion may cause the blade to fracture.
11. The first thick section is a 'waste cut' to provide a well-defined surface parallel to each additional cut. After the initial waste cut, raise the swivel arm and move the chuck towards the blade 5 mm using the positioning dial. Further thick sections (~1 mm) for microscopy can further be collected with this method.
    NOTE: In order to save valuable tissue, the waste cut can be omitted. When sectioning a sample with an uneven edge, however, it is critical that the peak of the specimen be lined up tangentially to the edge of the coring drill bit.
    1. Be sure to account for the kerf of the blade when sectioning. For example, to get a 5 mm section from a blade that has a kerf of 0.5 mm, move the sample and chuck 5.5 mm towards the blade.
12. After sectioning is complete, place the glass slide with the mounted specimen on a hot plate to melt the thermal epoxy. This allows for swift removal of bone blocks from the slide.
    NOTE: The protocol can be paused here.

3. Sample coring

1. Mount 5 mm bone sections to the bottom of a shallow aluminum tin (~8 cm in diameter) using the thermal epoxy bonding technique as described in steps 2.2-2.4.
2. Place tin on an XY machine table of the mill-drill press (see Table of Materials) and hand tighten fixturing clamps (Figure 4).
3. Insert 2 mm inner-diameter hollow-shaft jeweler's diamond-tipped coring drill bit (see Table of Materials) to the mill-drill chuck. Adjust the depth limiter to prevent coring through the tin (Figure 5).
4. Align the central anterior aspect of the bone sample beneath the drill bit while avoiding close contact with either the periosteum, endosteum, or highly trabecularized areas.
   NOTE: Automated selection of mid-anterior femoral cortices is not feasible as cortical thickness varies among individuals, especially with increasing age.
5. Fill the tin with distilled water to completely cover the sample. This prevents heat build-up, burning of the sample, and/or damage to the drill bit during coring.
   NOTE: To assess the possibility of heat damage caused by coring, an infrared thermometer was used to achieve temperature readings from the distilled water as the coring bit first penetrated the bones' surface. The temperature varied by 1 °C, from 22.9 – 23.9 °C among the ten samples cored for this test. Thus, we argue that heat-induced damage is negligible.
6. For the first few instances of contact between the core bit and bone, apply gentle pressure in order to wear a ring on the superior surface of the bone. This prevents deflection of the drill bit at the beginning of the coring process and ensures correct placement of the bit.
7. During coring, lift the drill bit in and out of the sample while keeping the bit's tip beneath the water’s surface. Continue this technique every few seconds to flush out trapped bone dust and ensure debris is not occluding the drill bit.
   NOTE: If the core is forming a conical shape, it is likely due to 1) allowing insufficient time to flush bone dust from the coring bit, and 2) coring is occurring too quickly. Increased speed may break off large pieces from the sample and pulverize the superior aspect.
8. After coring is complete, the resulting bone core may become lodged in the hollow-stemmed drill bit (Figure 6). Use a pair of fine-tipped forceps or a small Allen wrench to dislodge the core from the bit (Figure 2).
9. Store the cored sample in a labelled microcentrifuge tube in a cool and dry location until imaging.

4. Image processing routines for evaluating bone microarchitectural parameters from cortical bone cores

1. Reconstruction of µCT images
   1. Download and install the latest NRecon version at https://www.bruker.com/products/microtomography.html for reconstruction of the SRµCT projection images.
   2. Select the NRecon shortcut on the Desktop and the associated GPUReconServer will appear.
   3. Open desired dataset in pop-up window. If window does not appear, select the folder icon on the upper left-hand corner of the dataviewer window.
   4. Select the first projection from the SRµCT acquisition. Under Output, remove the selections for Use ROI and Scales ON.
   5. Choose the reconstruction file destination. Select Browse and create a new folder named Recon. The selected file format should be BMP(8).
   6. Check Misalignment Compensation.
      NOTE: This estimation is often close to correct. The rough 3D render can be manually adjusted by moving the arrows up and down to shift the overlapping images so that the right and left edges align as closely as possible.
   7. Under Settings, choose desired selections to apply Smoothing, Beam Hardening, CS Rotation, Object Larger than FOV, and Ring Artifacts algorithms.
   8. Adjust the histogram under Output by selecting Auto.
      NOTE: The resulting image may be dim.
   9. Select Start to begin processing the reconstruction.
   10. Use standard nomenclature for canal/osteocyte lacunar indices²⁸. These may include: total VOI tissue volume (TV), canal volume (Ca.V), total number of canals (Ca.N), average canal diameter (Ca.Dm), cortical porosity (Ca.V/TV), given as a percentage, total number of lacunae (N.Lc), and average lacunar volume (Lc.V), among others. To determine lacunar density per mm³ (N.Lc/BV), bone volume (BV) is calculated as total volume minus canal volume (TV-Ca.V).
       NOTE: The protocol can be paused here.
2. Collection of Microarchitectural Data from Reconstructed Images
   1. Download and install the latest version of CTAnalyser at https://www.bruker.com/products/microtomography/micro-ct-software/3dsuite.html for analysis of microarchitectural parameters.
      NOTE: The free version of CTAnalyser is limited in functionality. Therefore, it is recommended to purchase a full license to perform more detailed analyses.
   2. Under Image | Properties | Change Pixel size, ensure the pixel size matches that of the applied µCT imaging protocol.
      NOTE: If editing images in ImageJ or a similar program, note that upon saving, the header embedded in the TIFF file will be modified and the analysis software will alter the pixel size when importing the dataset.
   3. Select Custom Processing in order to create a task list (see Supplementary Materials) to analyze the bone microarchitecture from the scan dataset. A general protocol for osteocyte lacunar network parameters using plugins proprietary to CTAnalyser follows here:
      NOTE: The plugin task list works well for datasets where the sample is the only subject visible in the FOV. If empty space surrounds the specimen, the application of an ROI is required. Otherwise, the values gathered in 3D Analysis and Individual Object Analysis will be artificially decreased.
      1. Reload the images to reset and/or adjust any modifications (e.g., from editing in ImageJ or similar) prior to opening the Custom Processing menu in the analysis software.
      2. To reduce noise in the images, apply a Gaussian low-pass filter in 3D space with a round kernel and a radius of 2-3.
         NOTE: These settings were applied to datasets from the reported SRµCT experiments through trial and error testing. The goal was to obtain the best quality reconstructions for the data. Adjust reconstruction settings to suit each unique experimental setup.
      3. Apply a global grayscale threshold to the images by selecting low and high values to highlight vascular canals. The reconstructed slices seen in Figures 8B and 8D depict an example threshold of 0-155.
         NOTE: Similar to step 4.2.3.2, the threshold settings applied here were chosen through extensive trial and error. Thresholding should be adjusted for each experimental set-up and µCT imaging system used.
      4. Despeckle (denoise) to remove white speckles in 3D space that are within the volumetric pixel (voxel) size range of osteocyte lacunae in order to isolate canals only.
         NOTE: For a SRµCT scan of human cortical bone taken at 0.9 µm pixel size, the lower limit for osteocyte lacunae is 13 voxels.
      5. Despeckle to remove any black speckles in 2D space to remove artifacts in the canals. These can be quite large in 2D, thus remove features that are <15,000 pixels.
      6. Dilate the pores in 3D space using the Morphological operation function with a round kernel of 2 or 3 radius, depending on quality of images, in order to isolate any soft tissues trapped in canals.
      7. Perform an additional Despeckle function using the same settings as step 4.2.3.5. in order to remove isolated soft tissues within canals.
      8. Erode the dilation from step 4.2.3.6. using a Morphological operation function using a round kernel with either 2 or 3 radius. The radius for this step must match the radius used in Procedure 4.2.3.6.
      9. Run 3D Analysis and select which parameters to calculate for the volume of vascular canals. Generally, the basic values will provide sufficient information.
      10. Save the processed images with Save Bitmaps into a custom subfolder in the directory.
          NOTE: If creating a 3D reconstruction image from the processed images using a program such as Amira/Avizo, Dragonfly, Drishti, etc., saving the images as monochrome (1 bit) is recommended.
      11. Calculate the number of vascular canals and describe their size, shape, and orientation using the Individual Object Analysis function.
      12. Repeat steps 4.2.3.1 – 4.2.3.3. to reset the image for osteocyte lacunar analysis.
      13. Remove white speckles in 3D space using the Despeckle function, ensuring that such artifacts are smaller than the lower limit of lacunar size. This step removes noise from the scan that may appear to be cortical pores, while preserving true osteocyte lacunae. For human SRµCT scans of 0.9 µm pixel size, this lower limit is 13 voxels.
      14. Despeckle once more to remove white speckles that are larger than the upper limit of lacunar size. For human SRµCT datasets with the settings listed in step 4.2.3.13, this limit is 2743 voxels.
      15. Perform 3D Analysis to extract microstructural information pertaining to the osteocyte lacunae specifically.
      16. Select Save Bitmaps to save the processed images in order to isolate the osteocyte lacunae.
      17. Perform Individual Object Analysis to calculate the number of osteocytes in 3D within the selected Volume of Interest (VOI).
          NOTE: Once the task list has been established and tested, CTAnalyser has a batch manager (BatMan) function which can be employed to accelerate data extraction and ensure uniform image processing. A task list with example settings for Procedure 4.2.3. can be found in the Supplementary Materials.

Wyniki

The described method of core sampling proved to be highly effective and efficient. Coring specimens using this protocol allowed for the procurement of >300 consistently sized samples for experiments on the CLS BMIT-BM beamline², with an FOV of ~2 mm at 1.49 µm voxel size. To validate the consistency of core diameter, three measurements were taken along the length (top, middle, bottom) of a subset of human anterior femoral cores (n=69). The average diameter of the cores was 1.96 &...

Dyskusje

There has been no comprehensive, standardized protocol for procuring uniform and cylindrical cortical bone core samples for high-resolution SRµCT imaging with limited FOV setups. The protocol detailed here fills that void by providing a comprehensive tutorial regarding how to procure consistently sized cortical bone core samples for SRµCT imaging and the subsequent accurate visualization and extraction of microarchitectural data. We have shown that our protocol provides a more standardized and reliable met...

Materiały

Name	Company	Catalog Number	Comments
1-1/8" plunge cutting carbide for composites	Warrior	61812	28.6mm plunge
70% Ethanol	Fisher Scientific	BP8201500	3.8 Liters
Blunt-tipped forceps	Fisher Scientific	10-300
Centrifuge tubes	ThermoFisher	55398
Crystalbond 509-3 Epoxy	Ted Pella	821-3
CTAnalyser	Bruker microCT	v.1.15.4.0	Download and install at https://www.bruker.com/products/microtomography/micro-ct-software/3dsuite.html
Dental Tool Kit	Amazon	787269885110
Diamond wafering saw blade for composite material	Buehler	#11-4247
Drill Press	Jet Mill/Drill	350017	Model: JMD-15, benchtop drill presses are suitable substites, but typically lack a translatable machine table for positioning samples beneath the drill stem
Fine-tipped forceps	Fisher Scientific	22-327379
Fixturing clamps for XY machine table for mill/drill	MSC Industrial Supply	#04804571
Glass microscope slides	Ted Pella	26005	75x50mm slides, 1mm thick
Glass slide chuck	Buehler	#112488	Large enough to hold 75x50mm glass slides
Hot plate capable of reaching 140 °C	ThermoScientific	HP88850105
Incubator	NAPCO	Model 4200
Isocut Fluid	Buehler	111193032	Lubricant; 30mL
Jeweler's diamond coring drill bit	Otto Frei	#119.050	2mm inner diameter hollow stem coring bit
NRecon	Bruker microCT	v.1.6.10.2	Download and install at https://www.bruker.com/products/microtomography.html
Oscillating saw	Harbor Freight	62866
Oven-safe glass dishes	Pyrex	1117715	Glass food storage container
Precision slow-speed saw (Isomet 1000)	Buehler	111280160
Razor blades	Amazon	25181
Shallow aluminum tins	Amazon	B01MRWLD0R	~8cm diameter
Specimen cups	Amazon	616784425436 885334344729
Tergazyme detergent	Alconox	1304-1	1.8kg box
Ultrasonic cleaner	MTI Corporation	KJ201508006