Automating three-dimensional osteoarthritis histopathological grading of human osteochondral tissue using machine learning on contrast-enhanced micro-computed tomography.

OBJECTIVE: To develop and validate a machine learning (ML) approach for automatic three-dimensional (3D) histopathological grading of osteochondral samples imaged with contrast-enhanced micro-computed tomography (CEµCT). DESIGN: A total of 79 osteochondral cores from 24 total knee arthroplasty patients and two asymptomatic donors were imaged using CEµCT with phosphotungstic acid -staining. Volumes-of-interest (VOI) in surface (SZ), deep (DZ) and calcified (CZ) zones were extracted depth-wise and subjected to dimensionally reduced Local Binary Pattern -textural feature analysis. Regularized linear and logistic regression (LR) models were trained zone-wise against the manually assessed semi-quantitative histopathological CEµCT grades (diameter = 2 mm samples). Models were validated using nested leave-one-out cross-validation and an independent test set (4 mm samples). The performance was primarily assessed using Mean Squared Error (MSE) and Average Precision (AP, confidence intervals are given in square brackets). RESULTS: Highest performance on cross-validation was observed for SZ, both on linear regression (MSE = 0.49, 0.69 and 0.71 for SZ, DZ and CZ, respectively) and LR (AP = 0.9 [0.77-0.99], 0.46 [0.28-0.67] and 0.65 [0.41-0.85] for SZ, DZ and CZ, respectively). The test set evaluations yielded increased MSE on all zones. For LR, the performance was also best for the SZ (AP = 0.85 [0.73-0.93], 0.82 [0.70-0.92] and 0.8 [0.67-0.9], for SZ, DZ and CZ, respectively). CONCLUSION: We present the first ML-based automatic 3D histopathological osteoarthritis (OA) grading method which also adequately perform on grading unseen data, especially in SZ. After further development, the method could potentially be applied by OA researchers since the grading software and all source codes are publicly available.

Effect of centrifugal force on the development of articular neocartilage with bovine primary chondrocytes.

A lot has been invested into understanding how to assemble cartilage tissue in vitro and various designs have been developed to manufacture cartilage tissue with native-like biological properties. So far, no satisfactory design has been presented. Bovine primary chondrocytes are used to self-assemble scaffold-free constructs to investigate whether mechanical loading by centrifugal force would be useful in manufacturing cartilage tissue in vitro. Six million chondrocytes were laid on top of defatted bone disks placed inside an agarose well in 50-ml culture tubes. The constructs were centrifuged once or three times per day for 15 min at a centrifugal force of 771×g for up to 4 weeks. Control samples were cultured under the same conditions without exposure to centrifugation. The samples were analysed by (immuno)histochemistry, Fourier transform infrared imaging, micro-computed tomography, biochemical and gene expression analyses. Biomechanical testing was also performed. The centrifuged tissues had a more even surface covering a larger area of the bone disk. Fourier transform infrared imaging analysis indicated a higher concentration of collagen in the top and bottom edges in some of the centrifuged samples. Glycosaminoglycan contents increased along the culture, while collagen content remained at a rather constant level. Aggrecan and procollagen α1(II) gene expression levels had no significant differences, while procollagen α2(I) levels were increased significantly. Biomechanical analyses did not reveal remarkable changes. The centrifugation regimes lead to more uniform tissue constructs, whereas improved biological properties of the native tissue could not be obtained by centrifugation.

3D morphometric analysis of calcified cartilage properties using micro-computed tomography.

OBJECTIVE: Our aim is to establish methods for quantifying morphometric properties of calcified cartilage (CC) from micro-computed tomography (µCT). Furthermore, we evaluated the feasibility of these methods in investigating relationships between osteoarthritis (OA), tidemark surface morphology and open subchondral channels (OSCCs). METHOD: Samples (n = 15) used in this study were harvested from human lateral tibial plateau (n = 8). Conventional roughness and parameters assessing local 3-dimensional (3D) surface variations were used to quantify the surface morphology of the CC. Subchondral channel properties (percentage, density, size) were also calculated. As a reference, histological sections were evaluated using Histopathological osteoarthritis grading (OARSI) and thickness of CC and subchondral bone (SCB) was quantified. RESULTS: OARSI grade correlated with a decrease in local 3D variations of the tidemark surface (amount of different surface patterns (rs = -0.600, P = 0.018), entropy of patterns (EP) (rs = -0.648, P = 0.018), homogeneity index (HI) (rs = 0.555, P = 0.032)) and tidemark roughness (TMR) (rs = -0.579, P = 0.024). Amount of different patterns (ADP) and EP associated with channel area fraction (CAF) (rp = 0.876, P < 0.0001; rp = 0.665, P = 0.007, respectively) and channel density (CD) (rp = 0.680, P = 0.011; rp = 0.582, P = 0.023, respectively). TMR was associated with CAF (rp = 0.926, P < 0.0001) and average channel size (rp = 0.574, P = 0.025). CC topography differed statistically significantly in early OA vs healthy samples. CONCLUSION: We introduced a µ-CT image method to quantify 3D CC topography and perforations through CC. CC topography was associated with OARSI grade and OSCC properties; this suggests that the established methods can detect topographical changes in tidemark and CC perforations associated with OA.

Electromechanical properties of human osteoarthritic and asymptomatic articular cartilage are sensitive and early detectors of degeneration.

OBJECTIVE: To evaluate cross-correlations of ex vivo electromechanical properties with cartilage and subchondral bone plate thickness, as well as their sensitivity and specificity regarding early cartilage degeneration in human tibial plateau. METHOD: Six pairs of tibial plateaus were assessed ex vivo using an electromechanical probe (Arthro-BST) which measures a quantitative parameter (QP) reflecting articular cartilage compression-induced streaming potentials. Cartilage thickness was then measured with an automated thickness mapping technique using Mach-1 multiaxial mechanical tester. Subsequently, a visual assessment was performed by an experienced orthopedic surgeon using the International Cartilage Repair Society (ICRS) grading system. Each tibial plateau was finally evaluated with µCT scanner to determine the subchondral-bone plate thickness over the entire surface. RESULTS: Cross-correlations between assessments decreased with increasing degeneration level. Moreover, electromechanical QP and subchondral-bone plate thickness increased strongly with ICRS grade (ρ = 0.86 and ρ = 0.54 respectively), while cartilage thickness slightly increased (ρ = 0.27). Sensitivity and specificity analysis revealed that the electromechanical QP is the most performant to distinguish between different early degeneration stages, followed by subchondral-bone plate thickness and then cartilage thickness. Lastly, effect sizes of cartilage and subchondral-bone properties were established to evaluate whether cartilage or bone showed the most noticeable changes between normal (ICRS 0) and each early degenerative stage. Thus, the effect sizes of cartilage electromechanical QP were almost twice those of the subchondral-bone plate thickness, indicating greater sensitivity of electromechanical measurements to detect early osteoarthritis. CONCLUSION: The potential of electromechanical properties for the diagnosis of early human cartilage degeneration was highlighted and supported by cartilage thickness and µCT assessments.

In vitro method for 3D morphometry of human articular cartilage chondrons based on micro-computed tomography.

OBJECTIVE: The aims of this study were: to 1) develop a novel sample processing protocol to visualize human articular cartilage (AC) chondrons using micro-computed tomography (µCT), 2) develop and validate an algorithm to quantify the chondron morphology in 3D, and 3) compare the differences in chondron morphology between intact and osteoarthritic AC. METHOD: The developed protocol is based on the dehydration of samples with hexamethyldisilazane (HMDS), followed by imaging with a desktop µCT. Chondron density and depth, as well as volume and sphericity, were calculated in 3D with a custom-made and validated algorithm employing semi-automatic chondron selection and segmentation. The quantitative parameters were analyzed at three AC depth zones (zone 1: 0-10%; zone 2: 10-40%; zone 3: 40-100%) and grouped by the OARSI histological grades (OARSI grades 0-1.0, n = 6; OARSI grades 3.0-3.5, n = 6). RESULTS: After semi-automatic chondron selection and segmentation, 1510 chondrons were approved for 3D morphometric analyses. The chondrons especially in the deeper tissue (zones 2 and 3) were significantly larger (P < 0.001) and less spherical (P < 0.001), respectively, in the OARSI grade 3-3.5 group compared to the OARSI grade 0-1.0 group. No statistically significant difference in chondron density between the OARSI grade groups was observed at different depths. CONCLUSION: We have developed a novel sample processing protocol for chondron imaging in 3D, as well as a high-throughput algorithm to semi-automatically quantify chondron/chondrocyte 3D morphology in AC. Our results also suggest that 3D chondron morphology is affected by the progression of osteoarthritis (OA).

3D histopathological grading of osteochondral tissue using contrast-enhanced micro-computed tomography.

OBJECTIVE: Histopathological grading of osteochondral (OC) tissue is widely used in osteoarthritis (OA) research, and it is relatively common in post-surgery in vitro diagnostics. However, relying on thin tissue section, this approach includes a number of limitations, such as: (1) destructiveness, (2) sample processing artefacts, (3) 2D section does not represent spatial 3D structure and composition of the tissue, and (4) the final outcome is subjective. To overcome these limitations, we recently developed a contrast-enhanced µCT (CEµCT) imaging technique to visualize the collagenous extracellular matrix (ECM) of articular cartilage (AC). In the present study, we demonstrate that histopathological scoring of OC tissue from CEµCT is feasible. Moreover, we establish a new, semi-quantitative OA µCT grading system for OC tissue. RESULTS: Pathological features were clearly visualized in AC and subchondral bone (SB) with µCT and verified with histology, as demonstrated with image atlases. Comparison of histopathological grades (OARSI or severity (0-3)) across the characterization approaches, CEµCT and histology, excellent (0.92, 95% CI = [0.84, 0.96], n = 30) or fair (0.50, 95% CI = [0.16, 0.74], n = 27) intra-class correlations (ICC), respectively. A new µCT grading system was successfully established which achieved an excellent cross-method (µCT vs histology) reader-to-reader intra-class correlation (0.78, 95% CI = [0.58, 0.89], n = 27). CONCLUSIONS: We demonstrated that histopathological information relevant to OA can reliably be obtained from CEµCT images. This new grading system could be used as a reference for 3D imaging and analysis techniques intended for volumetric evaluation of OA pathology in research and clinical applications.

Determining collagen distribution in articular cartilage using contrast-enhanced micro-computed tomography.

OBJECTIVE: Collagen distribution within articular cartilage (AC) is typically evaluated from histological sections, e.g., using collagen staining and light microscopy (LM). Unfortunately, all techniques based on histological sections are time-consuming, destructive, and without extraordinary effort, limited to two dimensions. This study investigates whether phosphotungstic acid (PTA) and phosphomolybdic acid (PMA), two collagen-specific markers and X-ray absorbers, could (1) produce contrast for AC X-ray imaging or (2) be used to detect collagen distribution within AC. METHOD: We labeled equine AC samples with PTA or PMA and imaged them with micro-computed tomography (micro-CT) at pre-defined time points 0, 18, 36, 54, 72, 90, 180, 270 h during staining. The micro-CT image intensity was compared with collagen distributions obtained with a reference technique, i.e., Fourier-transform infrared imaging (FTIRI). The labeling time and contrast agent producing highest association (Pearson correlation, Bland-Altman analysis) between FTIRI collagen distribution and micro-CT -determined PTA distribution was selected for human AC. RESULTS: Both, PTA and PMA labeling permitted visualization of AC features using micro-CT in non-calcified cartilage. After labeling the samples for 36 h in PTA, the spatial distribution of X-ray attenuation correlated highly with the collagen distribution determined by FTIRI in both equine (mean ± S.D. of the Pearson correlation coefficients, r = 0.96 ± 0.03, n = 12) and human AC (r = 0.82 ± 0.15, n = 4). CONCLUSIONS: PTA-induced X-ray attenuation is a potential marker for non-destructive detection of AC collagen distributions in 3D. This approach opens new possibilities in development of non-destructive 3D histopathological techniques for characterization of OA.

Quantifying Subresolution 3D Morphology of Bone with Clinical Computed Tomography.

The aim of this study was to quantify sub-resolution trabecular bone morphometrics, which are also related to osteoarthritis (OA), from clinical resolution cone beam computed tomography (CBCT). Samples (n = 53) were harvested from human tibiae (N = 4) and femora (N = 7). Grey-level co-occurrence matrix (GLCM) texture and histogram-based parameters were calculated from CBCT imaged trabecular bone data, and compared with the morphometric parameters quantified from micro-computed tomography. As a reference for OA severity, histological sections were subjected to OARSI histopathological grading. GLCM and histogram parameters were correlated to bone morphometrics and OARSI individually. Furthermore, a statistical model of combined GLCM/histogram parameters was generated to estimate the bone morphometrics. Several individual histogram and GLCM parameters had strong associations with various bone morphometrics (|r| > 0.7). The most prominent correlation was observed between the histogram mean and bone volume fraction (r = 0.907). The statistical model combining GLCM and histogram-parameters resulted in even better association with bone volume fraction determined from CBCT data (adjusted R2 change = 0.047). Histopathology showed mainly moderate associations with bone morphometrics (|r| > 0.4). In conclusion, we demonstrated that GLCM- and histogram-based parameters from CBCT imaged trabecular bone (ex vivo) are associated with sub-resolution morphometrics. Our results suggest that sub-resolution morphometrics can be estimated from clinical CBCT images, associations becoming even stronger when combining histogram and GLCM-based parameters.