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.

The prognostic value of haemodynamic parameters in the recovery phase of an exercise test. The Finnish Cardiovascular Study.

We tested the hypothesis that the change from the peak to recovery values of systolic arterial pressure (SAP recovery) and rate-pressure product (RPP recovery) can be used to predict all-cause and cardiovascular mortality, as well as sudden cardiac death (SCD) in patients referred to a clinical exercise stress test. As a part of the Finnish Cardiovascular Study (FINCAVAS), consecutive patients (n=2029; mean age+/-SD=57+/-13 years; 1290 men and 739 women) with a clinically indicated exercise test using a bicycle ergometer were included in the present study. Capacities of attenuated SAP recovery, RPP recovery and heart rate recovery (HRR) to stratify the risk of death were estimated. During a follow-up (mean+/-s.d.) of 47+/-13 months, 122 patients died; 58 of the deaths were cardiovascular and 33 were SCD. In Cox regression analysis after adjustment for the peak level of the variable under assessment, age, sex, use of beta-blockers, previous myocardial infarction and other common coronary risk factors, the hazard ratio of the continuous variable RPP recovery (in units 1000 mm Hg x b.p.m.) was 0.85 (95% CI: 0.73-0.98) for SCD, 0.87 (0.78-0.97) for cardiovascular mortality, and 0.87 (0.81 to 0.94) for all-cause mortality. SAP recovery was not a predictor of mortality. The relative risks of having HRR below 18 b.p.m., a widely used cutoff point, were as follows: for SCD 1.28 (0.59-2.81, ns), for cardiovascular mortality 2.39 (1.34-4.26) and for all-cause mortality 2.40 (1.61-3.58). In conclusion, as a readily available parameter, RPP recovery is a promising candidate for a prognostic marker.