Ensemble Machine Learning Models For Oral Lesion Classification Using A Clinically Curated Histopathological Dataset

Read full text Summarise with VIRAT
Authors: H C Hemashree, Pradeep N, Ahmed Mujib B.R
Cited by: 0
References: 19

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Oral cancer, with the most common type being Oral Squamous Cell Carcinoma (OSCC), is a pervasive world-wide health problem whose successful treatment relies on early detection. In this paper, presents a novel framework of ensemble learning method for oral lesion classification. Starting with a clinically relevant dataset of 2450 histopathological images from six lesion classes, expert-annotated histopathological features were employed to train and refine three ensemble-based models: Random Forest (RF), XGBoost, and CatBoost. The methods show state-of-the-art performance, with CatBoost obtaining the highest test accuracy of 95% compared to XGBoost (94%) and Random Forest (92%). More importantly, it outperforms state-of-the-art methods in oral histopathology categorization and can compare favorably with advanced ensemble method for other cancer diagnostics. This work primary contribution is the integration of clinical domain knowledge with machine learning, taking advantage of explainable histopathological features to provide a solution both highly accurate and feasible in practice. This framework offers the automated detection of oral lesions at scale, using a reliable detection method at a lower cost, and can be particularly beneficial for clinics with fewer resources.

Fields of Study & Subjects

AI in cancer detection Oral Health Pathology and Treatment Artificial Intelligence in Healthcare

Publication details

Journal / source
International Journal of Advances in Signal and Image Sciences
Volume
12
Issue
3s
Pages
737–755
Publication date
1 January 2026
Language
EN