Quantitative analysis of optical coherence tomography and histopathology images of normal and dysplastic oral mucosal tissues

摘要

Selecting the most representative site for biopsy is crucial in establishing a definitive diagnosis of oral epithelial dysplasia. The current process involves clinical examination that can be subjective and prone to sampling errors. The aim of this study was therefore to investigate the use of optical coherence tomography (OCT) for differentiation of normal and dysplastic oral epithelial samples, with a view to developing an objective and reproducible approach for biopsy site selection. Biopsy samples from patients with fibro-epithelial polyps (n = 13), mild dysplasia (n = 2), and moderate/severe dysplasia (n = 4) were scanned at 5-μm intervals using an OCT microscope and subsequently processed and stained with hematoxylin and eosin (H&E). Epithelial differentiation was measured from the rate of change (gradient) of the backscattered light intensity in the OCT signal as a function of depth. This parameter is directly related to the density of optical scattering from the cell nuclei. OCT images of normal oral epithelium showed a clear delineation of the mucosal layers observed in the matching histology. However, OCT images of oral dysplasia did not clearly identify the individual mucosal layers because of the increased density of abnormal cell nuclei, which impeded light penetration. Quantitative analysis on 2D-OCT and histology images differentiated dysplasia from normal control samples. Similar analysis on 3D-OCT datasets resulted in the reclassification of biopsy samples into the normal/mild and moderate/severe groups. Quantitative differentiation of normal and dysplastic lesions using OCT offers a non-invasive objective approach for localizing the most representative site to biopsy, particularly in oral lesions with similar clinical features.