Anatomy-based algorithms for detecting oral cancer using reflectance and fluorescence spectroscopy

Area of Science:

• Biomedical Optics
• Oral Pathology
• Medical Spectroscopy

Background:

• Oral lesions require accurate diagnosis to differentiate benign conditions from potentially malignant or cancerous ones.
• Current diagnostic methods may be invasive or lack quantitative precision.

Purpose of the Study:

• To develop and validate noninvasive spectroscopic methods for distinguishing benign from dysplastic/malignant oral lesions.
• To create diagnostic algorithms that account for anatomical variations in spectral properties.

Main Methods:

• Collected in vivo reflectance and fluorescence spectra from 71 patients with oral lesions.
• Biopsied and histopathologically evaluated tissue specimens.
• Developed site-specific diagnostic algorithms based on spectral data.

Main Results:

• Individual site algorithms achieved higher diagnostic accuracy (ROC-AUC 0.75 for tongue) than combined-site algorithms (ROC-AUC 0.60).
• Algorithms for sites with similar spectral properties (e.g., floor of mouth and tongue) showed good performance (ROC-AUC 0.71).

Conclusions:

• Spectroscopic detection of oral disease necessitates accounting for anatomical site variations.
• Anatomy-based algorithms demonstrate superior diagnostic performance for oral lesion characterization compared to generalized approaches.