# Harnessing the power of ANN for early detection and prediction of oral cancer

## TL;DR

This study shows that artificial neural networks can accurately detect oral cancer using optical refractive index data, offering a fast and reliable early screening method.

## Contribution

The study introduces an optimized ANN model using optical refractive index features for early oral cancer detection with high precision and specificity.

## Key findings

- The optimized ANN model achieved 98.72% precision in identifying cancerous cases.
- The model demonstrated 99.00% specificity in correctly identifying non-cancerous cases.
- ANN-assisted optical analysis shows potential for efficient and reliable early oral cancer screening.

## Abstract

Oral cancer affects millions of people worldwide, and early detection significantly improves treatment outcomes and survival rates. Conventional diagnostic approaches often face challenges related to subjectivity and delayed identification. In this context, artificial intelligence–based tools offer promising opportunities for rapid and reliable early screening.

This study investigates the feasibility of using an Artificial Neural Network (ANN) to predict oral cancer risk based on optical refractive index (RI) features. RI data corresponding to reported INOK (normal oral cells) and YD-10B (oral cancer cells) cell lines were employed. To enhance model robustness and assess feasibility, the dataset was synthetically augmented. Multiple ANN architectures and hyperparameter configurations were systematically evaluated to identify the optimal network topology for classification.

The optimized ANN model demonstrated excellent performance in distinguishing between normal and oral cancer cell data. A precision score of 98.72% indicates that nearly all samples classified as cancerous were truly positive, minimizing false-positive predictions. Additionally, the model achieved a specificity of 99.00%, highlighting its strong capability to correctly identify non-cancerous cases.

The high precision and specificity values underscore the effectiveness of ANN-based classification using optical refractive index features for oral cancer screening. By reducing false positives and preventing unnecessary anxiety among healthy individuals, the proposed approach offers significant clinical value. These findings demonstrate the potential of ANN-assisted optical analysis as a reliable and efficient tool for early oral cancer detection, paving the way for faster diagnosis and improved patient outcomes.