DenseNet121-Powered Histopathological Image Analysis for Accurate Lung Cancer Detection

Lung cancer is one of the most common causes of cancer-related fatalities worldwide, and its impact on healthcare systems is growing. While histopathological analysis is still the gold standard for diagnosis, it is subjective and time-consuming due to its heavy reliance on pathologists’ knowledge. Deep learning methods have been popular in medical imaging due to the need for automated, effective, and precise diagnostic tools. In this work, deep learning models for automatic histopathology image categorization are investigated, including VGG19, InceptionV3, ResNet, ViT, and DenseNet121. 15,000 photos from the Kaggle “Lung and Colon Cancer Histopathological Images” dataset, which covers several histological subtypes of lung cancer, make up the dataset employed. To increase the generalization of the model, data preprocessing includes resizing, augmentation, and standardization. The models are assessed using a variety of performance metrics, including accuracy, precision, recall, F1 score, confusion matrix, and ROC curves. They are trained using the Adam optimizer with a learning rate of 1e-4. The experimental findings show that deep learning models can detect lung cancer with high accuracy, indicating their potential to help pathologists make an accurate and timely diagnosis.