Intelligent immune clonal optimization algorithm for pulmonary nodule classification.

Area of Science:

• Medical Imaging and Diagnostics
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Computer-aided diagnosis (CAD) systems are crucial for early lung cancer detection via pulmonary nodule classification.
• Existing CAD systems suffer from low accuracy and high false-positive rates (FPR) in pulmonary nodule classification.
• Addressing these limitations is vital for improving diagnostic efficacy.

Purpose of the Study:

• To develop a novel intelligent immune clonal selection and classification algorithm for enhanced pulmonary nodule classification.
• To overcome the limitations of low accuracy and high FPR in current CAD systems.
• To improve the early detection of lung cancers through more precise nodule classification.

Main Methods:

• A novel method utilizing chaotic motion with logistic mapping to generate a diverse initial population for the immune algorithm.
• Development of an intelligent mutation strategy incorporating Gaussian mutation operator (GMO) and Cauchy mutation operator (CMO) for a Gauss-Cauchy hybrid mutation operator.
• Application of the proposed intelligent immune clonal optimization algorithm for pulmonary nodule classification.

Main Results:

• The proposed method achieved a high accuracy of 97.87% in pulmonary nodule classification.
• The algorithm demonstrated a low false-positive rate (FPR) of 1.52 false positives per scan (FPs/scan).
• Experimental validation on 90 CT scans with 652 nodules confirmed the method's effectiveness.

Conclusions:

• The intelligent immune clonal optimization algorithm significantly enhances pulmonary nodule classification accuracy.
• The developed method effectively reduces the false-positive rate in computer-aided diagnosis systems.
• This approach shows great promise for improving early lung cancer detection and CAD system performance.