Early Detection and Classification of Lung Cancer using Segment Anything Model 2 and Dense Net

Lung cancer is one of the most perilous diseases worldwide with high incidence and low survival rates due to late diagnosis. Accurate detection and diagnosis of lung nodules is important for early-stage detection. Machine learning and deep learning techniques have greatly improved the precision of lung nodule segmentation and classification in Computed Tomography (CT) images. The study presents a novel approach to segmenting and classifying nodules by combining foundational models with deep learning architectures. We have used the Segment Anything Model (SAM2) to segment lung nodules and DenseNet to classify them as benign and malignant. SAM2 has been tested on the datasets using different prompts to achieve better results. Foundational Models and Deep Learning architecture’s integration significantly improved Computer-Aided Detection (CADe) and Computer-Aided Diagnosis (CADx) in medical images. Experimental results proved the effectiveness of the proposed model for early-stage detection and classification of lung nodules from CT scans. SAM2 model achieves a Dice Similarity Coefficient (DSC) of 97.87% and an Intersection over Union (IoU) of 95.82% for segmentation, and the DenseNet model's classification accuracy is 97.34%. The experimental results demonstrate the performance of our model compared to existing techniques.