If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Propagation phase-contrast micro-computed tomography allows laboratory-based three-dimensional imaging of articular cartilage down to the cellular level
High-resolution non-invasive three-dimensional (3D) imaging of chondrocytes in articular cartilage remains elusive. The aim of this study was to explore whether laboratory micro-computed tomography (micro-CT) permits imaging cells within articular cartilage.
Design
Bovine osteochondral plugs were prepared four ways: in phosphate-buffered saline (PBS) or 70% ethanol (EtOH), both with or without phosphotungstic acid (PTA) staining. Specimens were imaged with micro-CT following two protocols: 1) absorption contrast (AC) imaging 2) propagation phase-contrast (PPC) imaging. All samples were scanned in liquid. The contrast to noise ratio (C/N) of cellular features quantified scan quality and were statistically analysed. Cellular features resolved by micro-CT were validated by standard histology.
Results
The highest quality images were obtained using propagation phase-contrast imaging and PTA-staining in 70% EtOH. Cellular features were also visualised when stained in PBS and unstained in EtOH. Under all conditions PPC resulted in greater contrast than AC (p < 0.0001 to p = 0.038). Simultaneous imaging of cartilage and subchondral bone did not impede image quality. Corresponding features were located in both histology and micro-CT and followed the same distribution with similar density and roundness values.
Conclusions
Three-dimensional visualisation and quantification of the chondrocyte population within articular cartilage can be achieved across a field of view of several millimetres using laboratory-based micro-CT. The ability to map chondrocytes in 3D opens possibilities for research in fields from skeletal development through to medical device design and treatment of cartilage degeneration.
. There is potential for micro-computed tomography (micro-CT) to reveal full-field 3D information from intact segments of articular cartilage. Success could advance the study of osteoarthritis development and provide a natural pattern to enable full-field strain measurements of cartilage
. Articular cartilage is typically difficult to image by micro-CT due to poor absorption of X-rays owing to its composition of low-Z elements such as carbon, hydrogen, oxygen and nitrogen
Phase contrast imaging x-ray computed tomography: quantitative characterization of human patellar cartilage matrix with topological and geometrical features.
Contrast agents, typically composed of high-density heavy metals, bind to extracellular components, increasing density and thereby increase the resulting signal
. Pauwels et al., conducted an investigation of 28 potential contrast agents and found that the most promising contrast agents for soft-tissue staining were phosphomolybdic acid (H3PMo12O40, PMA), phosphotungstic acid (H3PW12O40, PTA) and mercury chloride (HgCl2)
. These changes in water content likely modify the mechanical properties of the material and thus may not be fully representative of the in vivo conditions. Imaging samples immersed in alternative liquids, altering less the physiological conditions, would be attractive.
Phase-contrast imaging detects the phase shift that X-rays experience passing through matter, resulting from differences in refractive index or non-uniform thicknesses
Imaging applications of synchrotron x-ray phase-contrast microtomography in biological morphology and biomaterials science. I. General aspects of the technique and its advantages in the analysis of millimetre-sized arthropod structure.
. Many of these methods, including both analyser-based diffraction enhanced contrast and grating interferometry, require specialised and modified equipment. PPC requires no such modification to the equipment used for typical absorption contrast as long as the beam is highly spatially coherent and the scanner allows for the detector to be positioned sufficiently far from the sample. Various attempts have been made to image articular cartilage with phase-contrast using relatively common laboratory micro-CT
Phase contrast imaging x-ray computed tomography: quantitative characterization of human patellar cartilage matrix with topological and geometrical features.
. The ability to image the cartilage structure in lab micro-CT would be highly advantageous, enabling wider access for researchers in utilising technology which may eventually allow in vivo diagnostics with clinically-available tomography.
Herein we explore the hypothesis that a combination of staining and laboratory phase-contrast micro-CT can adequately visualise individual chondrocytes within intact samples of articular cartilage. Our primary aim was to develop a method to image chondrocytes with laboratory micro-CT. Our secondary aim was to achieve this whilst maintaining the tissue in conditions which deviate the least from those found physiologically.
Method
Sample preparation
Fresh juvenile bovine (<2 years old) stifle joints (n = 3) from two animals were obtained from a slaughterhouse and stored at −25°C until experimentation. The specimens were thawed at 4°C, kept hydrated with Dulbecco's phosphate buffered saline solution (DPBS, #14190-094, Fisher Scientific, USA) and 3 mm cylindrical osteochondral plugs (n = 10) were taken from the femoral condyles using a hollow punch. Samples were prepared under four conditions (Fig. 1). All were initially bathed for 30 min in phosphate-buffered saline (PBS) and the ethanol (EtOH) treated samples were immersed in step-wise increasing concentrations of EtOH
Decreased solute adsorption onto cracked surfaces of mechanically injured articular cartilage: towards the design of cartilage-specific functional contrast agents.
. From each of the PBS or EtOH treated groups, one osteochondral plug was maintained in the liquid without staining (denoted PBS or EtOH), whilst others (Table I and supplementary table) were further processed with staining using a 1% w/v phosphotungstic acid solution (weight/volume, H3PW12O40, PTA, #79690 Sigma–Aldrich, USA) in either PBS or 70% ethanol for 21 h then rinsed in the medium prior to micro-CT scanning (denoted PBS + PTA or EtOH + PTA). The full description of each sample is included in the supplementary table. Staining time was optimised in pilot experiments. All samples were stored in their liquid medium at room temperature prior to scanning.
Fig. 1Sample preparation methodology for the four preparation methods using osteochondral plugs extracted from a similar position in the bovine condyles.
Table IMicro-CT scan parameters. All scans took place at 40 kV and 75 μA. AC = Absorption contrast; SOD = Source-to-Object Distance; ODD = Object-to-Detector Distance. No X-ray filters were used. The supplementary table provides a further breakdown of parameters for each individual sample and includes sources for each sample
Samples were immersed in their corresponding liquid medium (PBS or 70% EtOH) and mounted in sealed plastic containers. All scans were carried out on a Versa 520 X-ray micro-CT scanner (Zeiss, Germany). For all scans, voltage and current were 40 kV and 75 μA respectively with no pixel binning. No X-ray filters were applied. To allow comparison of the signal between absorption and propagation phase signals, scans under different scanning protocols were taken. Image quality at different source-to-object (SOD) and object-to-detector (ODD) distances can be maintained by adjusting exposure time to ensure a sufficient photon count reaches the detector. The larger the SOD and ODD distances, the longer the exposure time required to maintain the photon count. Firstly the absorption signal was collected by minimising the SOD and ODD distances to reduce scanning time. This is the type of scan typically carried out with a micro-CT scanner. Next the phase-contrast scan was taken with enlarged SOD and ODD, allowing implementation of PPC. Using a larger SOD also has the advantage of decreasing cone beam error
. Increasing the ODD for the PPC scan reduced the X-ray flux, resulting in exposure times typically four times longer than for the absorption scan. The full set of scan parameters for each scan are shown in Table I and in the supplementary table. A volume of interest of approximately 2x2x2 mm was included for each set of conditions. For the bovine samples used during this study, the volume of interest did not constitute the entire cartilage height. For histological comparison, an additional sample (e) was prepared by the same method employed for the EtOH + PTA sample (d) and scanned to maximise image quality with a higher number of projections and further increased SOD and ODD. Reconstruction of the projection images to produce 3D volumetric data sets was performed using the Reconstructor Scout-and-Scan software (Zeiss, Germany). The reconstructed CT volumes were visualized and analysed using (Fiji Is Just) ImageJ software
(version 1.52g, NIH, USA). Two of the sample from groups PBS + PTA (b) and EtOH + PTA (d) were scanned twice using PPC to ensure scan repeatability and measure consistency in scan quality.
Histology
Following the micro-CT scan, one of the osteochondral plugs (e) was stored in 10% neutral buffered formalin (#HT501128, Sigma–Aldrich) for 24 h. Before paraffin wax embedding, PTA was removed by ion-exchanging in a washing solution of 0.55mM NaOH, 0.1 M of Na2HPO4, 137 mM NaCl, and 2.7mM KCl, pH 10 for 5 days following established protocol
, paraffin-embedded and sections (5 μm) collected on Superfrost slides (Fisher Scientific, USA). Sections were dewaxed immersing twice for 5 min in Gentaclear (Genta Medical, UK), washed in tap water and subsequently stained with: (1) Alcian Blue at pH 2.5 with counterstaining of nuclei with Neutral Red
. All sections slides were cover slipped with DPX mountant (Sigma–Aldrich, USA). Colour micrographs were acquired using a Zeiss Axio Observer Inverted Widefield Microscope with an IC5 colour camera, and with a fully motorised stage controlled by ZEN Blue pro software capable of tiling and stitching, using a 20 × DIC Plan Apochromat air objective with numerical aperture of 0.8, and 2048 × 2048 resolution, with pixel size corresponding to 0.33 μm.
Image analysis
Contrast to noise analysis
Of the constituent components within articular cartilage, the largest individual features are chondrocytes. Yet the chondrocytes have previously been difficult to visualise over large distances in 3D
, owing to their small size, low spatial density and previously discussed low contrast. Typically each chondrocyte will occupy few voxels and noise can easily lead to erroneous segmentation. Therefore, as a measure of contrast and scan quality, the contrast to noise ratio (C/N) of individual chondrocyte features was calculated through the sample height for each sample and compared between the samples scanned under different conditions. Within each sample [Fig. 2(A)] a region was micro-CT scanned [Fig. 2(B)]. Eleven equally spaced layers per sample from the reconstructed image stack [Fig. 2(C)] were analysed in ImageJ using the Plot Profile tool [Fig. 2(D) & (E)] with a line length of 60 pixels to measure grayscale intensity across cells and their surrounding matrix (n = 10 per layer, n = 110 cells per sample)
. Chondrocytes were visualised with a higher grayscale value than the surrounding matrix. A higher C/N indicates more clearly visualised features. Care was taken to only include one chondrocyte per profile therefore only the highest peak characterised a cell, all others represented the surrounding matrix. A MATLAB script (R2016a, MathWorks Inc, Natick, MA, USA) computed the ratio of the amplitude of the maximum peak (A) above the background peaks for each plot, divided by the standard deviation of the surrounding noise peaks
Fig. 2Methodology to extract contrast to noise (C/N) values in the micro-CT scans of osteochondral plugs. A volume of interest within the cartilage region of the osteochondral plug (A.) was scanned, as shown by the black dashed cube (B.). To measure changes through the cartilage height, eleven layers equally spaced 100 layers apart (≈200 μm; C.) were taken from within the 1000-layer thick z-stack and on each of these layers ten features were analysed with the Surface Plot Tool in ImageJ (denoted by dashed lines; D.) to extract the grayscale intensity as a function of distance from the cell (E.), resulting in 110 plots per sample. Greyscale values were processed in MATLAB: the amplitude of the signal (“A”) was divided by the standard deviation of the noise (“N”) to give the C/N. Images shown are for sample (d.1) EtOH + PTA with PPC.
The micro-CT and histological images for sample (e) were manually registered in ImageJ. Similarly located areas of 500 μm × 500 μm were selected in both modalities (micro-CT n = 12, histology n = 3 for each of the four stains) at three different heights through the sample (n = 36 in total each for micro-CT and histology). Three consecutive micro-CT slices were combined using the Max Intensity Z project process within Fiji ImageJ software to result in a comparable thickness (5.91 μm) to that of the histological slices (5 μm). The histological images were scaled to the same resolution as the micro-CT. Images were segmented using the Trainable Weka Segmentation
and circularity of 0.15–1.00 to measure density and roundness.
Statistical analysis
Contrast to noise ratio (C/N) data for each micro-CT scan was imported to SPSS (IBM SPSS Statistics, version 25, Armonk, NY). Students t-tests were run to understand the effects of liquid medium (PBS or EtOH), staining (with or without PTA) and scan-type (absorption and propagation imaging) on C/N, and between the two scan-types for each preparation technique (Table I (a)-(d)). Data are mean ± standard deviation unless otherwise stated. Contrast to noise values were normally distributed, as assessed by Q–Q plot inspection. By inspection of boxplots, 14 of the 20 groups contained a limited number of outliers further than 1.5 box lengths. These constituted at most 1.8% of all values per group and thus were included. There was non-homogeneity of variances for groups (p < 0.0005) as assessed by Levene's test for equality of variances therefore the Welch's t-test was chosen, for which equal variances are not assumed. Paired-sample t-tests were run on the C/N values for the two pairs of repeat scans of samples prepared with PBS + PTA and EtoH + PTA. There were no outliers in the PBS + PTA groups and two significant outliers were found in the EtOH + PTA data (comprised of n = 220 data points) which were not excluded. Assessment by Q–Q plot inspection showed normally distributed differences in C/N scores for both conditions.
Results
PPC imaging following PTA staining in EtOH yields greatest chondrocyte visualisation
Our data show that the most efficient combination of imaging modality, sample preparation medium and staining to resolve cellular details in articular cartilage was to use PPC scanning of samples stained with PTA in EtOH [Fig. 3(D)]. Cellular features in cartilage were also visualised in samples stained with PTA prepared in PBS as well as in unstained samples stored in EtOH. For both scan types, C/N for samples in ethanol were higher than PBS (p < 0.0001, Fig. 3(A)), and higher for samples stained in PTA compared to unstained samples (p < 0.0001, Fig. 3(B)). The C/N score for ethanol was higher than PBS for both unstained (p = 0.001) and stained samples (p < 0.0001, Fig. 3(C)). Propagation scanning increased C/N for all sample preparation methods [Fig. 3(D)]. There was no difference between the pairs of repeat scans in either the PBS + PTA or EtOH + PTA samples (p = 0.852 and p = 0.112 respectively, Fig. 3(E)), showing scan repeatability. Simultaneous imaging of subchondral bone and cartilage did not impede image quality.
Fig. 3Contrast to noise (C/N) data, as described in Fig. 2, was statistically analysed with t-tests in SPSS and displayed are mean contrast to noise ratio for the three variables: scan type (absorption and propagation), medium (PBS and EtOH) and staining (none and PTA-staining) with their 95% CIs. (A.) Between mediums for the two scan protocols C/N was greater for EtOH than the PBS in both scanning methods (both p < 0.0001) with mean differences of 1.243 (95% CI 0.709 to 1.778) for absorption contrast and 6.231 (95% CI 5.772 to 6.690) for propagation contrast. (B.) Between unstained and unstained samples for the two scan protocols. For absorption contrast, staining increased C/N by a mean value of 2.325 (p < 0.0001, 95% CI 1.824 to 2.837) and by 7.202 (p < 0.0001, 95% CI 6.786 to 7.618) for the propagation method. (C.) Between mediums with and without staining. Unstained, there was an increase in the C/N of 0.524 (p = 0.001, 95% CI 0.220 to 0.828) using ethanol rather than PBS and when using PTA staining there was an increase of 6.279 (p < 0.0001, 95% CI 5.83 to 6.730). (D.) Mean contrast to noise values for all scans divided by preparation technique. PPC increase C/N for all groups compared to AC: (a) PBS by 0.400 (p < 0.038, 95% CI (0.022–0.779), (b) PBS + PTA by 2.490 (p < 0.0001, 95% CI 2.093 to 2.885), (c) EtOH by 3.654 (p < 0.0001, 95% CI 2.782 to 4.527) and (d) by 8.224 (p < 0.0001, 95% CI 7.686 to 8.762). (E.) Two repeat propagation scans were taken of samples from the PTA-stained groups. No difference in mean C/N was found with either medium: PBS had a mean difference of 0.058 (p = 0.852, 95% CI -0.560 to 0.676), EtOH had a mean difference of 1.183 (p = 0.112, 95% CI 0.281 to 2.648).*P < 0.05, **P < 0.001, ***P < 0.0001.
Contrast to noise ratio values through the height of each sample are shown in Fig. 4. A control was included as a measure of contrast variation in the surrounding matrix which was similar in both scanning methods. For most of the sample preparation techniques contrast improved using the propagation phase scanning method [Fig. 4(B)] compared to the absorption scans [Fig. 4(A)] and remained improved through most of the sample height.
Fig. 4Mean contrast to noise ratio (C/N) values (with 95% CIs) through the height of osteochondral plugs prepared with various methods and micro-CT scanned using the two Absorption Contrast (A.) and Propagation Phase (B.) protocols. Chondrocyte grayscale intensity (n = 10) was plotted across 11 scan layers per sample as described in Fig. 2. The scans included a depth of approximately 2 mm of articular cartilage mid-way through its height. 100 micro-CT slices equates to approximately 200 μm. For comparison, data collected without cellular features is given as a control for signal variation in the surrounding matrix.
Reconstructed z-slice images from the micro-CT scans for each sample preparation method (Fig. 5) infer the same results as the C/N values suggest: that propagation phase-contrast shows more clearly defined features [Fig. 5(B)]. Features are most clearly visualised with staining in the ethanol group. Features reduce in size and intensity throughout the sample height.
Fig. 5Comparison of image quality through the height of samples under the different sample preparation techniques and micro-CT protocols. Layer 1 is in the cartilage above the subchondral bone, and layer 950 is distally located closer to the articular surface. For illustrative purposes brightness was normalised between sets of images, all analysis was carried out on unedited images.
In order to validate our findings from micro-CT analysis, we performed image registration between micro-CT and conventional cartilage histology images of the same sample (Fig. 6 and supplementary video). Both techniques allowed observation and quantification of similar cellular features (Fig. 6(B), (D), (E)) and cellular distribution Fig. 6(C)). We attempted to correlate cellular area between the two imaging methods, but the area of an individual cell was too sensitive to greyscale thresholding for this to be accurately feasible. However, the density of cellular features observed with micro-CT was between the values for chondrocytes and their lacunae in the histological images [Fig. 6(C)] and cellular roundness was similar [Fig. 6(D)].
Fig. 6Micro-CT comparison with histology. (A.) Reconstructed micro-CT volume and light microscopy (LM) histological slices (Alcian Blue, Masson's Trichrome; Picro-Sirius Red and Safranin O) of sample (e): 70% EtOH and PTA staining. The scale of the micro-CT and histology images is comparable. (B.) A high magnification region of interest between the two techniques is shown with symbols denoting corresponding features. Equivalent 500 × 500 μm areas were analysed between both methods (each n = 36) and mean values (with 95% CIs) are shown for cellular density (C.) and cellular roundness (D.). Measurements for lacunae encompass the chondrocytes and other cellular features.
Depth-dependent feature analysis (n = 12) was carried out between micro-CT and histology (Fig. 7). Density of cellular features reduced with distance from the subchondral bone for both methods (Fig. 7(A), (B), (C)). Roundness remained similar for both micro-CT and histology throughout the sample height [Fig. 7(D)].
Fig. 7Reconstructed micro-CT slice of 1.97 μm thickness of osteochondral plug sample (e): 70% EtOH and PTA staining (A.), with higher magnification volume renderings of different regions through the height of the cartilage: deep, middle and superior (B.). Renderings are of a volume of 135 x 135 × 80 μm produced with Fiji's Volume Viewer. Measures of density (C.), and roundness (D.) were calculated over 500 × 500 μm areas at the three locations for both micro-CT and histology (n = 12). Measurements for lacunae encompass the chondrocytes and other cellular features. Displayed values are mean and 95% CIs.
We report for the utility of a standard laboratory micro-CT scanner to visualise and quantify features of the chondrocyte population within intact articular cartilage in 3D. Histological staining was used to confirm these cartilaginous features observed by micro-CT at the cellular level. Images between both methods were successfully registered, confirming the location and distribution of features. Measurements of cellular density measured with micro-CT yielded values within the range of chondrocytes and their lacunae measured with histological images. Morphology was compared with cellular roundness, for which both techniques yielded similar values. Repeatability of measured C/N values was confirmed for both PBS and EtOH-stained samples. Imaging was successful using propagation phase-contrast imaging with the sample maintained within a liquid environment and is compatible with either PBS or EtOH as a medium, achieving the aims of the study. It is pertinent also that we find that simultaneous imaging of hard and soft tissues did not impede image quality. Propagation phase-contrast increases the contrast of individual chondrocytes compared to using absorption contrast. This offers researchers the opportunity to image chondrocyte distributions in 3D without specialised synchrotron equipment, enabling investigations such as chondrocyte morphology across grades of cartilage damage, 3D strain mapping techniques such as digital volume correlation to evaluate mechanical properties in situ, and models for 3D finite element analysis in silico simulations.
This study represents a complimentary addition to the growing body of evidence supporting the non-destructive imaging of the constituents of articular cartilage. Previous studies have differed in focus on other aspects of the cartilage structure
. We have compared and quantified the output scans of samples prepared using different preparation techniques and scanning signals. As with previous studies it was found that heavy metal staining provided an improvement in signal attenuation
. The use of PBS as the medium during sample preparation and subsequent staining is atypical in previous literature and provides a more physiological environment than EtOH or formalin fixation. The visualised features in this study are comparable to those achieved for similarly prepared samples in micro-CT and synchrotron facilities using a similar voxel size
, and additionally can image the adjacent subchondral bone. For sample (e) which was processed with EtOH and PTA staining we found the cellular density of chondrocyte features to reduce from 663 mm−2 to 511 mm−2 when approaching the cartilage surface. An earlier study using confocal microscopy and sectioning of bovine samples found cellular density to follow a similar trend with the lowest density furthest from the articular surface
. We note that there was a discrepancy in density between the chondrocytes and their lacunae. A potential cause of this was damage incurred by sectioning and associated processing, this is avoided with non-destructive visualisation techniques such as micro-CT. Micro-CT values for roundness were approximately 7% higher than with histology. This may be due to partial volume effect artefacts observed with micro-CT scanning
Our study has several limitations. Host tissue was stored frozen at −25°C and thawed for use. Successful histological staining of nuclei and cells post-scanning suggest that tissue disruption is not any more extensive than would be expected in samples prepared in this way. Previous studies have reported no difference in mechanical properties between fresh and frozen soft tissues
yet testing the method works with fresh tissue would be beneficial. Currently, this method has only been applied to a small number of juvenile bovine samples, and future studies are needed to increase the sample size and to confirm that the method works with human articular cartilage. Penetration issues were experienced by virtue of the low-energy X-rays being easily absorbed by the sample and surrounding container, and the method is limited by specimen size. It was found that the thickness of the sample container had an effect on the signal reaching the detector. Wall thickness was kept below 1 mm to reduce weakening of the signal. Given that samples were scanned within liquid, using containers of large internal diameter decreased the detected signal due to increased liquid volume. The largest samples we have scanned with this technique were 6 mm in diameter. This provided sufficient resolving power at the perimeters and centre of the sample but suffered from inconsistent signal quality in the intermediate region. The scans presented in this study include ≈2 mm of the bovine cartilage height in the volume of interest. Ideally the whole cartilage thickness would have been included yet owing to the large thickness of bovine cartilage we sought to preserve resolution where possible. For the absorption scans a pixel size of 3.5 μm was used, compared to 2 μm for the phase contrast scans. Ideally this variable would have been removed but limitations with the scanner required a larger pixel size. Scanning parameters could not be kept consistent between scans owing to differences in sample density due to the different preparation techniques. Attempts were made to keep the overall scanning time similar for all samples but differences in exposure time and number of projections may still have affected comparison between the scans. Previous studies have shown that ionizing radiation can have a significant effect on a range of measured properties in articular cartilage samples
. Low energy X-rays interact with these low-density materials and cause more damage than high energy beams causing particular problems for the low voltages used throughout this study
. These effects could be reduced by limiting the number of projections, and therefore scanning time. Moini et al., have reported colour changes in amino acids upon irradiation
, and we observed that some of the samples stored in EtOH and stained with PTA had a temporary blue hue after scanning. Further work is necessary to determine whether these observations have negative implications for this mode of imaging. Currently the method has been validated in 2D against histological sections, further work is recommended for validation against established 3D techniques such as confocal imaging.
Herein, we report a novel and validated non-destructive technique to visualise chondrocyte features through a region of several millimetres in articular cartilage. This enables an objective quantification of chondrocyte distribution and morphology in three dimensions allowing greater insight for investigations into studies of cartilage development, degeneration and repair. One such application of our method, is as a means to provide a 3D pattern in the cartilage which, when combined with digital volume correlation, could determine 3D strain gradient measurements enabling potential treatment and repair of cartilage degeneration. Moreover, the method proposed here will allow evaluation of cartilage implanted with tissue engineered scaffolds designed to promote chondral repair, providing valuable insight into the induced regenerative process.
Author contributions
Conceived and designed the experiments: JNC, JRTJ, UH.
Performed the experiments: JNC, AG, SAF.
Analysis and interpretation of the data: JNC, SAF, BJ, AAP, SMR, JRTJ, UH.
Drafting of the article: JNC, JRTJ, UH.
Critical revision of the article for important intellectual content: JNC, AG, SAF, BJ, AAP, SMR, JRTJ, UH.
Final approval of the article: JNC, AG, SAF, BJ, AAP, SMR, JRTJ, UH.
The authors gratefully acknowledge the financial support of EPSRC (Engineering and Physical Sciences Research Council) funding, United Kingdom (EP/N025059/1 and EP/K027549/1) and the first author holds the Imperial College Class of 1964 Scholarship, United Kingdom. The authors (BJ and AAP) are also indebted to Versus Arthritis, United Kingdom (grant no. 20581) for their support. Beyond financial support, the funding sources had no involvement in the preparation of this manuscript.
Acknowledgements
The authors wish to thank Maria Parkes for providing specimens, Farah Ahmed and Brett Clark for their assistance with the micro-CT imaging, Lorraine Lawrence for histological sectioning and staining and Kiron Athwal for assistance with the statistical analysis.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Phase contrast imaging x-ray computed tomography: quantitative characterization of human patellar cartilage matrix with topological and geometrical features.
Imaging applications of synchrotron x-ray phase-contrast microtomography in biological morphology and biomaterials science. I. General aspects of the technique and its advantages in the analysis of millimetre-sized arthropod structure.
Decreased solute adsorption onto cracked surfaces of mechanically injured articular cartilage: towards the design of cartilage-specific functional contrast agents.
31242-7&id=gr1.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}
31242-7&id=gr2.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}
31242-7&id=gr3.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}
31242-7&id=gr4.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}
31242-7&id=gr5.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}
31242-7&id=gr6.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}
31242-7&id=gr7.jpg&_x_tr_sl=en&_x_tr_tl=iw&_x_tr_hl=iw&_x_tr_pto=sc){kind=link}