Diverse spectral band-based deep residual network for tongue squamous cell carcinoma classification using fiber optic Raman spectroscopy.

Abstract The research is to propose a new classification framework, called diverse spectral band-based deep residual network (DSB-ResNet), which can distinguish tongue squamous cell carcinoma (TSCC) from non-cancerous tissue. A fiber optic Raman spectroscopy system is used to collect Raman spectral data of TSCC and normal tissues. DSB-ResNet takes advantage of diverse spectral band-based spectra without processing to derive spectral representations from different spectral bands of Raman spectra, which improves the ability to identify TSCC. To show the superiority of the proposed method, the existing methods are used as the competitive methods to compare with the DSB-RestNet, the results demonstrate our method has the highest performance with 97.38%, 98.75%, and 98.25% for sensitivity, specificity, and accuracy, respectively. The experimental results show that the DSB-ResNet is able to distinguish TSCC from non-cancerous tissue successfully. The proposed method is expected to provide a theoretical and methodological base for accurate detection of TSCC.