Dual Path Feature Extraction

DualPathTransformer is a deep learning framework for lung cancer classification from chest CT images. The model combines dual-path CNN-based local and global feature extraction with Transformer-based attention fusion to effectively capture fine-grained details and global context.

Overview

Early detection of lung cancer is critical for improving survival rates. However, accurate classification from CT images remains challenging due to subtle visual patterns. This project proposes a hybrid CNN–Transformer architecture that models both local spatial details and global semantic context for robust lung cancer classification.

Model Architecture

• Local Path: ResNet18-based CNN for extracting fine-grained spatial features
• Global Path: ResNet18-based CNN for capturing global contextual information
• Fusion: Transformer Encoder with a learnable classification token
• Output Classes: Normal, Benign, Malignant

Dataset

• IQ-OTH/NCCD Lung CT Dataset
• There are three classes: Normal, Benign, Malignant

Results

• Validation Accuracy: 94%
• Improved performance compared to baseline CNN-based models
• Evaluation metrics include accuracy, precision, recall, F1-score, and ROC–AUC

Tech Stack

• Python
• PyTorch
• NumPy, Matplotlib, Seaborn
• Scikit learn