Accurate and Interpretable Oral Lichen Planus Classification Using a Hybrid Framework with Attention-Guided Feature Fusion and Hybrid Optimisation

Description

This study utilised a dataset of 1,346 clinical photographs comprising 747 OLP and 599 non-OLP cases collected from the College of Dental Medicine, Rangsit University. To address class imbalance and improve model generalisation, we applied targeted augmentation, expanding the dataset to 8,000 balanced images. The dataset was stratified into training, validation and testing subsets using an 80:10:10 split. Develop two custom convolutional neural networks (CNNs), LichenNet and Inception-LichenNet, to capture both fine-grained and high-level features. Deep features were extracted from both models and integrated using an attention-guided feature fusion strategy. To further optimise efficiency and reduce redundancy, we applied a hybrid feature selection (GWO + See Tree) optimisation. For clinical interpretability, Grad-CAM was used to generate class-discriminative visual explanations. All experiments were conducted on Kaggle GPU environments using 10-fold cross-validation to ensure reliable evaluation

Steps to reproduce

Setup Instructions Download and Extract • Download Lichen Planus Experiment.zip from this repository. • Unzip the file to your local machine. Load the Code on Kaggle • Navigate to the Full Experiment Code folder. • Upload Oral Lichen Planus.py to your Kaggle Notebook. Connect the Dataset • Add the required .npy feature files from All Features as Kaggle datasets. • Update dataset paths in the script to point to the uploaded Train, Validation, and Test feature/label files. Run the Models • Execute the notebook sequentially. The pipeline will perform: o Training and evaluation of LichenNet and Inception-LichenNet o Attention-guided feature fusion o Hybrid feature selection o Explainable AI (Grad-CAM visualizations) • Outputs include predictions, confusion matrices, and classification metrics. View Results • Open the Confusion Matric and Results folder. • Review: o Confusion matrix images (.png) o Performance reports (.csv) Requirements Environment • Kaggle Notebook (recommended) • Python ≥ 3.8 Libraries • TensorFlow / Keras • NumPy, Pandas, scikit-learn • Matplotlib, Seaborn Hardware • GPU: NVIDIA Tesla P100 or higher (recommended)

Institutions

• King Faisal University

  Eastern, Al-Hasa
• Chulalongkorn University Faculty Of Dentistry

  Bangkok
• Rangsit University

  Mueang Pathum Thani

Categories

Licence