Non-Small Cell Lung Cancer Classification Using Densely Connected Neural Network

Lung cancer varies in malignancy depending on the condition and arises when cells in the lungs grow uncontrollably and become malignant. There are two main types: Non-Small Cell Lung Cancer (NSCLC) and Small Cell Lung Cancer (SCLC). NSCLC, the most prevalent form, accounts for about 85% of lung cancer cases and includes subtypes like pulmonary adenocarcinoma (ADC) and pulmonary squamous cell carcinoma (SqCC). SCLC, though less common, grows faster and is more likely to metastasize at diagnosis. Current screening using low-dose Computed Tomography (CT) scans has limited sensitivity and a high false-positive rate, with over 90% of positive results being non-cancerous. Employing Convolutional Neural Network (CNN) with DenseNet architecture could enhance non-invasive diagnostic accuracy by classifying nodules based on texture, shape, and size.