Lung disease classification in chest X-ray images using optimal cross stage partial bidirectional long short term...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Deep Learning

Background:

• Lung disease poses a significant threat to respiratory health and survival.
• Current lung disease classification methods face challenges in accuracy, computational complexity, and overfitting.
• Deep learning offers a promising avenue to address these limitations in lung disease detection.

Purpose of the Study:

• To propose an advanced deep learning model for accurate classification of various lung diseases.
• To overcome the limitations of existing models in terms of detection accuracy and computational efficiency.
• To develop a robust system for identifying COVID, lung opacity, pneumonia, and normal lung conditions.

Main Methods:

• Image pre-processing including data augmentation, filtering, and resizing.
• Threshold-based segmentation to isolate relevant lung regions.
• Utilizing an Optimal Cross Stage Partial Bidirectional Long short term memory (OCBiNet) network for classification.
• Optimizing model parameters with the Improved Mother Optimization (ImMO) algorithm, incorporating Logistic Chaotic Mapping.

Main Results:

• The Improved Mother Optimization (ImMO) algorithm demonstrated enhanced convergence for global solution finding.
• The OCBiNet model achieved high performance metrics in lung disease classification.
• Evaluated metrics included Accuracy (99.11%), Recall (98.98%), Precision (99.18%), and F-Score (99.08%).

Conclusions:

• The proposed OCBiNet model demonstrates superior performance in lung disease classification.
• The integration of ImMO algorithm enhances the efficiency and accuracy of the deep learning model.
• The findings suggest a significant advancement in automated lung disease detection systems.