Machine Learning Classification of Articular Cartilage Integrity Using Near Infrared Spectroscopy

Isaac O Afara,Jaakko K Sarin,Simo Saarakkala,Rami K Korhonen,Juha Töyräs,Mikko A J Finnilä,Walter Herzog,Simo Ojanen

doi:10.1007/s12195-020-00612-5

Abstract

IntroductionAssessment of cartilage integrity during arthroscopy is limited by the subjective visual nature of the technique. To address this shortcoming in diagnostic evaluation of articular cartilage, near infrared spectroscopy (NIRS) has been proposed. In this study, we evaluated the capacity of NIRS, combined with machine learning techniques, to classify cartilage integrity.MethodsRabbit (n = 14) knee joints with artificial injury, induced via unilateral anterior cruciate ligament transection (ACLT), and the corresponding contra-lateral (CL) joints, including joints from separate non-operated control (CNTRL) animals (n = 8), were used. After sacrifice, NIR spectra (1000–2500 nm) were acquired from different anatomical locations of the joints (nTOTAL = 313: nCNTRL = 111, nCL = 97, nACLT = 105). Machine and deep learning methods (support vector machines–SVM, logistic regression–LR, and deep neural networks–DNN) were then used to develop models for classifying the samples based solely on their NIR spectra.ResultsThe results show that the model based on SVM is optimal of distinguishing between ACLT and CNTRL samples (ROC_AUC = 0.93, kappa = 0.86), LR is capable of distinguishing between CL and CNTRL samples (ROC_AUC = 0.91, kappa = 0.81), while DNN is optimal for discriminating between the different classes (multi-class classification, kappa = 0.48).ConclusionWe show that NIR spectroscopy, when combined with machine learning techniques, is capable of holistic assessment of cartilage integrity, with potential for accurately distinguishing between healthy and diseased cartilage.

Highlights

Assessment of cartilage integrity during arthroscopy is limited by the subjective visual nature of the technique
We compared the performance of traditional machine learning techniques, including support vector machines (SVM) and logistic regression (LR), with deep learning methods, deep neural networks (DNN), for classification of cartilage integrity based on near infrared spectroscopy (NIRS)
Histological analysis shows significant decrease in depth-wise PG content at the superficial and middle zones of the anterior cruciate ligament transection (ACLT) group compared to the CNTRL group, and throughout the cartilage depth compared to the CL group (Fig. 1h)

Summary

INTRODUCTION

Osteoarthritis (OA) is a disabling musculoskeletal condition affecting a significant proportion of the world’s population. Few studies have applied machine learning techniques, such as support vector machines (SVM)[3] and neural networks (ANN),[28,30] for analysis of cartilage NIR spectral data, regression analysis. Classification of cartilage integrity based on NIRS harnesses the capacity of the spectrum, which encodes latent and inherent properties of the cartilage matrix, to provide a holistic assessment of the tissue. We compared the performance of traditional machine learning techniques, including support vector machines (SVM) and logistic regression (LR), with deep learning methods, deep neural networks (DNN), for classification of cartilage integrity based on NIRS. We hypothesized that machine learning techniques, which encompass both traditional (e.g., SVM and LR) and state-of-the-art (e.g., DNN) artificial intelligence techniques, can harness sample-related information embedded in the NIR spectrum for classification and holistic assessment of the tissue integrity

MATERIALS AND METHODS

RESULTS

Findings

CONFLICT OF INTEREST