ECAM-Net: A Lightweight Convolutional Neural Network Enhanced with Efficient Channel Attention for Lung Cancer Classification

Lung cancer (LC) remains the primary cause of cancer-related deaths globally, underscoring the vital need for early detection to improve outcomes. While manual interpretation of computed tomography (CT) scans plays a key role in identifying lung cancer, it is often associated with errors and extended processing times. These limitations have driven the exploration of computational methods, especially deep learning networks, to facilitate faster and more accurate classification of various lung cancer types. This article proposed a lightweight convolutional neural network ECAM-Net (Efficient Channel Attention with MobileNet) using a pre-trained mobilenet and attention model to classify lung cancer images. The paper proposed a modified version of efficient channel attention mechanism (ECA) to improve feature representation. The Modified Efficient Channel Attention (ECA) consist multi kernel convolutional method for local cross-channel interaction to enhance its computational efficiency and performance. ECAM-Net presents an efficient classification model for lung cancer, demonstrating effective computational capabilities appropriate for mobile devices and edge computing platforms. The multi-class lung cancer dataset IQ-OTH/NCCD validates the performance of the proposed model. The experiments demonstrate that the proposed architecture achieved an accuracy of 99.35%, outperforming recent models.