Introduction

Assessment 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.

Methods

Rabbit (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 (n_TOTAL = 313: n_CNTRL = 111, n_CL = 97, n_ACLT = 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.

Results

The 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).

Conclusion

We 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.

中文翻译：

---

使用近红外光谱对关节软骨完整性进行机器学习分类。

介绍

关节镜检查过程中对软骨完整性的评估受到该技术的主观视觉性质的限制。为了解决关节软骨诊断评估中的这一缺点，已提出近红外光谱（NIRS）。在这项研究中，我们评估了 NIRS 结合机器学习技术对软骨完整性进行分类的能力。

方法

兔 ( n  = 14) 人工损伤膝关节，通过单侧前交叉韧带横断 (ACLT) 和相应的对侧 (CL) 关节诱导，包括来自单独的非手术对照 (CNTRL) 动物的关节 ( n  = 8 ), 被使用。牺牲后，从关节的不同解剖位置获取 NIR 光谱 (1000–2500 nm) ( n _TOTAL  = 313: n _CNTRL  = 111, n _CL  = 97, n _ACLT = 105)。然后使用机器和深度学习方法（支持向量机 - SVM、逻辑回归 - LR 和深度神经网络 - DNN）来开发模型，以仅根据样本的 NIR 光谱对样本进行分类。

结果

结果表明，基于SVM的模型在区分ACLT和CNTRL样本方面效果最佳（ROC_AUC = 0.93，kappa = 0.86），LR能够区分CL和CNTRL样本（ROC_AUC = 0.91，kappa = 0.81），而DNN是区分不同类别的最佳选择（多类别分类，kappa = 0.48）。

结论

我们表明，当 NIR 光谱与机器学习技术相结合时，能够对软骨完整性进行整体评估，并具有准确区分健康和患病软骨的潜力。