Oral Cancer Prediction Using a Probability Neural Network (PNN).

Area of Science:

• Oncology
• Computer Vision
• Machine Learning

Background:

• Oral cancer is often diagnosed at advanced stages in India, necessitating early detection methods.
• Identifying oral cancer early significantly improves patient prognosis and survival rates.
• Early detection of oral malignancy presents considerable challenges due to lesion heterogeneity.

Purpose of the Study:

• To develop and evaluate a computer-aided diagnostic tool for early oral cancer detection.
• To enhance the accuracy of oral malignancy prediction using advanced computational techniques.
• To address the challenges in identifying oral cancer at its incipient stages.

Main Methods:

• Utilized a probabilistic neural network (PNN) for oral malignancy prediction.
• Integrated discrete wavelet transform with PNN to improve cancer cell identification accuracy.
• Explored various computer vision techniques for analyzing oral lesions.

Main Results:

• The PNN model achieved a classification accuracy of 80% for predicting oral malignancy.
• The combination of PNN and discrete wavelet transform demonstrated effectiveness in accurate cancer cell prediction.
• Computer vision techniques were investigated to overcome challenges in identifying heterogeneous oral lesions.

Conclusions:

• Effective oral screening is vital for timely decision-making regarding oral lesions.
• Prompt referrals based on accurate screening can significantly reduce oral cancer mortality rates.
• The proposed PNN-based approach shows promise for early and accurate oral cancer detection.