HYBRID DEEP LEARNING MODEL FOR LUNG DISEASE PREDICTION

Abstract

 Lung cancer remains the global health concern because millions of people lost their lives in recent years. Traditional diagnostic method depends on manual CT scan interpretation that leads to variation in the diagnosis of lung cancer. This paper proposed advanced deep learning techniques for automated lung cancer prediction using CT scan images. The proposed model uses ResNet50 for the feature extraction and Convolutional Block Attention Module (CBAM) to enhance nodule detection by dynamically weighting critical spatial and channel wise features. This enhanced features then passed to multi-layer perception (MLP) for classification of the images. The proposed hybrid model achieved 98% accuracy for multi classification. The findings validate that hybrid deep learning architectures can significantly enhance accuracy in  lung cancer prediction in  the medical field.