Related Experiment Video For Artificial intelligence (AI)
Updated: Jan 11, 2026
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Development and validation of an AI-based application for early detection and risk stratification of oral potentially
Akash Gajanan Prabhune1, Vinay R Srihari1, Shreya Shree1
1ADMIRE Centre for Advancing Digital Health, Institute of Health Management Research Bangalore, India.
Background:
Oral Potentially Malignant Disorders (OPMDs) are early indicators of oral cancer, and timely detection is essential for improving patient outcomes. However, diagnosis often relies on expert clinical evaluation, which may not be available in low-resource settings.
Objective:
This study presents the development and validation of PRAYAAS, an AI-based mobile application for early detection and risk stratification of OPMDs using intraoral images.
Methods:
A total of 794 intraoral images were classified into three categories: (1) Normal mucosa/inflammatory conditions, (2) Premalignant conditions, and (3) Oral carcinoma. Images were split into training (70 %), validation (18 %), and test (12 %) datasets while maintaining class balance. Preprocessing involved resizing to 224 × 224 pixels, contrast enhancement, and normalization. A U-Net-based model segmented lesion regions, followed by classification using a fine-tuned DenseNet201 model. Model performance was evaluated using accuracy, precision, recall, F1-score, and confusion matrices.
Results:
The DenseNet201 classifier achieved 94 % accuracy on the test set. For normal/inflammatory lesions, precision and recall were 1.00. For premalignant lesions, precision was 0.87 and recall was 1.00. For carcinoma, precision was 1.00 and recall was 0.80. The integrated segmentation module improved lesion focus and reduced background noise. The app provides class-wise risk scores and a user-friendly interface for clinical support.
Conclusion:
PRAYAAS offers a robust, mobile-enabled solution for early OPMD screening. By integrating segmentation and classification into a single platform, the tool holds promise for enhancing community-based oral cancer detection and referral.
