An Evaluation of the Prognosticative Value of Hyalinization in the Biological Behaviour of Oral Lesions Using Image Analysis

OBJECTIVE: To evaluate the efficacy of artificial intelligence-based analysis of polychromatic staining in oral premalignant and malignant lesions and to predict their biological behavior. The study also aims to evaluate the prognostic value of collagen in these oral lesions. METHODS: In this study, a total of 45 histopathologically diagnosed normal (15), Premalignant lesions (n-15), and oral squamous cell carcinoma (n=15) were included. The tissue sections were subjected to routine Hematoxylin and Eosin (H and E) staining and a differential staining technique- Herovici’s stain. The stained slides were viewed under 10x magnification in the microscope and images of these sections were captured. The images were labelled, transferred and stored in the computer for image analysis. The collagen content in the photomicrographs were analysed using Image J software. The results were tabulated and subjected to Kruskal- Wallis test using the SPSS software. RESULTS: A significant increase in the amount of type III (blue stained) collagen fibers, compared to type I collagen fibers, was seen as the lesion progressed from premalignant disorders to oral squamous cell carcinoma. Normal mucosa showed predominantly type I (red stained) collagen fibers. The difference in the ratio of type I and III collagen fibers between the three groups was found to be statistically significant (P= 0.00). CONCLUSION: The study concluded that a significant change in stromal collagen composition exists, with an increase in the amount of type III collagen, that can be correlated to the lesion’s progression from premalignant to oral squamous cell carcinoma. Differential staining is an inexpensive and highly reproducible method of evaluating the composition of the stroma and using Image analysis to carry out this analysis makes the process more objective and renders it bias free.