Fig. 3.

The construction of PortNet. This figure illustrates the structural design of PortNet, highlighting its convolutional layers, feature extraction modules, and classification pathway. The network begins with a series of 1 × 1 convolution layers to reduce dimensionality and capture essential features, followed by a main processing block that combines feature maps. The final layers consist of average pooling, softmax activation, and output classification, enabling accurate and efficient lung cancer cell categorization.