Determination of the pathological staging of the oral squamous cell carcinoma based on microscopic hyperspectral...

This study proposes a novel methodology based on microscopic hyperspectral imaging (MHI) technology combined with machine learning for staging oral squamous cell carcinoma (OSCC), aiming to establish an optical method for quantitative pathological staging of OSCC.

In this study, the average spectrum of each histopathological section was extracted, and spectral differences between cross-sections were quantified by summing Euclidean distances. Subsequently, multiple preprocessing methods were applied to refine the raw spectral data. Principal component analysis (PCA) and the successive projections algorithm (SPA) were then employed for dimensionality reduction to optimize the classification model. Finally, extreme learning machine (ELM), support vector machine (SVM), and partial least squares discriminant analysis (PLS-DA) were utilized as classification models for staging prediction.

All three classification models demonstrated robust performance. Notably, the SPA-ELM model achieved optimal results, attaining 100% accuracy, sensitivity, and specificity irrespective of preprocessing. Furthermore, its classification precision was corroborated by a kappa statistic of 1.0 and near-perfect ROC curves.

The integration of MHI with machine learning presents a robust framework for clinicopathological staging of OSCC.