- Lung cancer is a widespread and deadly kind of cancer that has a high death rate globally. Conventional diagnostic techniques that rely on visual examination by healthcare professionals are susceptible to mistakes and are time-consuming because of the similarities between tumours. This paper presents a new strategy for classifying lung tumours using a deep feature fusion technique. At first, deep features are obtained from Computer Tomography (CT) images using various Convolutional Neural Network (CNN) structures such as ResNet18, ResNet50, Alex-Net, and DenseNet201. These features are classified using Support Vector Machines, K-Nearest Neighbors, Linear Discriminant, and Naïve Bayes. The highest-performing separate feature vectors, specifically ResNet50 and DenseNet201, are combined to construct a discriminative and informative fused feature vector. The proposed vector demonstrates a substantial increase in accuracy (95%) when compared to separate vectors. In addition, we assess the effectiveness of our strategy by comparing it to other advanced techniques, showcasing its resilience in classifying lung tumors. This approach shows potential for early illness diagnosis by healthcare experts.

Index Terms- Feature fusion, Deep learning, Machine learning, Nedical image diagnosis and Lung Cancer.