Computer-Aided Diagnosis system for diagnosis of pulmonary emphysema using bio-inspired algorithms

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonary Medicine

Background:

• Pulmonary emphysema involves lung alveoli destruction, causing shortness of breath and chronic cough.
• Accurate diagnosis of emphysema from chest CT scans is crucial for timely intervention.

Purpose of the Study:

• To design and implement a Computer-Aided Diagnosis (CAD) framework for detecting pulmonary emphysema in chest CT slices.
• To evaluate the efficacy of bio-inspired algorithms for feature selection in emphysema diagnosis.

Main Methods:

• Lung tissue segmentation and region of interest (ROI) extraction using Spatial Intuitionistic Fuzzy C-Means clustering.
• Feature extraction (shape, texture, run-length) from emphysematous lesions (centrilobular, paraseptal, bullae).
• Optimal feature subset selection using Moth-Flame Optimization (MFO) and Firefly Optimization (FFO) with Support Vector Machine (SVM) accuracy as fitness, followed by Extreme Learning Machine (ELM) classification.

Main Results:

• The CAD framework demonstrated high accuracy, precision, recall, and specificity in classifying emphysema presence.
• MFO and FFO algorithms outperformed other tested bio-inspired algorithms in feature selection for both real-time and public datasets.

Conclusions:

• The developed CAD framework effectively diagnoses pulmonary emphysema using chest CT images.
• Bio-inspired optimization algorithms, particularly MFO and FFO, are valuable tools for enhancing diagnostic accuracy in pulmonary emphysema detection.