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.

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.

Imaging of Osteoarthritic Human Articular Cartilage using Fourier Transform Infrared Microspectroscopy Combined with Multivariate and Univariate Analysis.

The changes in chemical composition of human articular cartilage (AC) caused by osteoarthritis (OA) were investigated using Fourier transform infrared microspectroscopy (FTIR-MS). We demonstrate the sensitivity of FTIR-MS for monitoring compositional changes that occur with OA progression. Twenty-eight AC samples from tibial plateaus were imaged with FTIR-MS. Hyperspectral images of all samples were combined for K-means clustering. Partial least squares regression (PLSR) analysis was used to compare the spectra with the OARSI grade (histopathological grading of OA). Furthermore, the amide I and the carbohydrate regions were used to estimate collagen and proteoglycan contents, respectively. Spectral peak at 1338 cm(-1) was used to estimate the integrity of the collagen network. The layered structure of AC was revealed using the carbohydrate region for clustering. Statistically significant correlation was observed between the OARSI grade and the collagen integrity in the superficial (r = -0.55) and the deep (r = -0.41) zones. Furthermore, PLSR models predicted the OARSI grade from the superficial (r = 0.94) and the deep (r = 0.77) regions of the AC with high accuracy. Obtained results suggest that quantitative and qualitative changes occur in the AC composition during OA progression, and these can be monitored by the use of FTIR-MS.

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.