A Method for Optimal Detection of Lung Cancer Based on Deep Learning Optimized by Marine Predators Algorithm.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Lung cancer is a leading cause of mortality, necessitating early detection for improved patient outcomes.
• Current diagnostic methods can be invasive or lack sufficient accuracy for early-stage identification.

Purpose of the Study:

• To develop and evaluate an optimal methodology for the early detection of lung cancer.
• To enhance diagnostic accuracy using a novel deep learning model integrated with metaheuristic optimization.

Main Methods:

• A new convolutional neural network (CNN) was designed for lung cancer detection.
• The Marine Predators Algorithm (MPA) was employed for optimizing the CNN architecture and improving performance.
• Image processing techniques were utilized in conjunction with deep learning models.
• The proposed method was validated on the RIDER dataset and compared against established deep networks (ResNet-18, GoogLeNet, AlexNet, VGG-19).

Main Results:

• The proposed MPA-based deep learning method demonstrated superior performance.
• Achieved high accuracy (93.4%), sensitivity (98.4%), and specificity (97.1%).
• Outperformed existing state-of-the-art methods, including pre-trained deep networks, with the lowest error rate (1.6%).

Conclusions:

• The developed methodology offers a highly efficient and accurate approach for early lung cancer detection.
• The integration of deep learning and metaheuristic optimization shows significant promise for improving cancer diagnostics.
• This MPA-based CNN method represents a significant advancement in non-invasive lung cancer screening.