Deep Hybrid Learning for Accurate Lung Nodule Classification in CT Scans

Lung cancer is one of the leading causes of cancer-related deaths worldwide, and early detection is critical for improving patient survival rates.This study presents an advanced CT scan-based image analysis pipeline for the reliable detection and classification of pulmonary nodules.The proposed method combines image preprocessing, segmentation, and feature extraction with deep learning classification to improve accuracy and robustness.We address challenges such as class imbalance, domain shift, and inter-class similarityparticularly between benign and normal cases-by applying targeted augmentation, class-balanced losses, and vesselsuppression techniques.Experimental evaluation on benchmark datasets shows that our approach achieves high accuracy and recall, especially for malignant cases, while minimizing false negatives in benign detection.The results highlight the potential of our method for integration into computer-aided diagnosis systems to support radiologists in clinical decision-making.