Residual-SwishNet: a deep learning-based approach for reliable lung cancer classification

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Lung cancer is a leading cause of global cancer deaths, necessitating improved diagnostic tools.
• Current computer-aided detection systems face challenges in feature extraction, accuracy, and generalizability.

Purpose of the Study:

• To develop a deep learning model for accurate and reliable lung cancer classification.
• To enhance feature extraction and classification performance for lung cancer detection.

Main Methods:

• A modified ResNet50 architecture, Residual-SwishNet, was developed.
• The ReLU activation function was replaced with Swish, and three dense layers were added for enriched feature representation.
• Softmax output with Cross-Entropy Loss was used to address class imbalance.

Main Results:

• Residual-SwishNet achieved high classification accuracies of 99.60% on LUNA16 and 99.11% on IQ-OTH/NCCD.
• The model demonstrated superior performance compared to existing state-of-the-art techniques.

Conclusions:

• The proposed Residual-SwishNet model shows significant potential as a robust tool for lung cancer diagnosis.
• This deep learning approach offers improved accuracy and reliability for early lung cancer detection.