Abstract—The use of deep learning for automatic classification has expanded rapidly, particularly through Convolutional Neural Network (CNN) architectures. AlexNet, a widely used CNN, offers a simple and stable structure. However, its performance on Computed Tomography (CT) scan images is limited by challenges in processing high-dimensional data and the scarcity of 3D CT scan datasets, often resulting in suboptimal classification outcomes. The Multi-Dimensional Slicing (MDS) method addresses this issue by converting 3D CT scan images into 2D representations along the axial, coronal, and sagittal planes. This technique enables comprehensive analysis of lung structures using more manageable datasets while preserving critical information. In this study, MDS is integrated with the ALVS-Net architecture, an improved version of AlexNet that incorporates a Visual Geometry Group (VGG) based structure in the final layer to reduce parameter complexity and enhance training stability. The Sigmoid-weighted Linear Unit (SiLU) activation function replaces the traditional Rectified Linear Unit (ReLU) function to improve nonlinear learning capabilities. The results of this study show that the proposed combination of MDS and ALVS-Net achieves excellent performance in classifying lung infections into four classes. This model achieves an accuracy, precision, recall, and F1−Score, all reaching 94%, a G-mean value of 93%, and a Cohen’s kappa value of 92%. These results suggest that the proposed method is effective and reliable for classifying lung infections based on CT scan images.
 
Keywords—lung infection, Computed Tomography (CT) scan, health risk, public health


Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).