Improved graph neural network-based green anaconda optimization for segmenting and classifying the lung cancer

Area of Science:

• Oncology
• Computer Science
• Medical Imaging

Background:

• Lung cancer arises from genetic damage to normal cells, with smoking as a primary cause.
• Environmental factors like asbestos, arsenic, and radon gas also contribute to lung cancer risk.
• Accurate and timely diagnosis is crucial for effective lung cancer treatment.

Purpose of the Study:

• To develop and evaluate a deep learning approach for accelerating lung cancer diagnosis.
• To improve the accuracy of classifying lung cancer subtypes using histopathology images.
• To optimize a novel deep learning model for enhanced diagnostic performance.

Main Methods:

• Utilized a benchmark dataset of lung cancer histopathology slides.
• Applied Gabor filter for image pre-processing and modified expectation maximization (MEM) for segmentation.
• Extracted features using the histogram of oriented gradient (HOG) scheme.
• Employed an improved graph neural network (IGNN) optimized by the green anaconda optimization (GAO) algorithm for classification.

Main Results:

• The proposed IGNN model achieved high accuracy in classifying lung cancer into normal, adenocarcinoma, and squamous cell carcinoma.
• Simulation findings demonstrated superior performance compared to existing diagnostic methods.
• Parameter optimization using GAO enhanced the overall accuracy maximization of the GNN.

Conclusions:

• The developed deep learning framework offers a promising tool for accurate and efficient lung cancer diagnosis.
• The IGNN model, optimized with GAO, shows significant potential for clinical application in histopathology analysis.
• This approach can aid in faster and more reliable identification of lung cancer subtypes.