Altruistic seagull optimization algorithm enables selection of radiomic features for predicting benign and malignant...

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Computational Biology

Background:

• Differentiating indeterminate pulmonary nodules is clinically challenging.
• Low-dose computed tomography (LDCT) generates extensive radiomic data for lung cancer screening.
• Accurate radiomic feature selection is crucial for reliable malignancy risk prediction.

Purpose of the Study:

• To propose the Altruistic Seagull Optimization Algorithm (AltSOA) for radiomic feature selection.
• To develop a pulmonary nodule malignancy risk prediction model using optimized features.
• To enhance the accuracy and efficiency of lung cancer screening methods.

Main Methods:

• Implemented AltSOA, incorporating altruism for global optimization.
• Designed a multi-objective fitness function for model training.
• Identified 11 key radiomic features, including Gray Level Co-occurrence Matrix (GLCM), using AltSOA.

Main Results:

• The AltSOA-selected features achieved a precise malignancy risk prediction (AUC = 0.8383).
• The model demonstrated superior performance compared to radiologists.
• The selected feature set was smaller, improving model efficiency.

Conclusions:

• AltSOA is a superior method for radiomic feature selection in pulmonary nodule analysis.
• The developed prediction model shows promise for improving lung cancer screening.
• The study highlights the potential of AI in enhancing diagnostic accuracy for pulmonary nodules.