ABOA-CNN: auction-based optimization algorithm with convolutional neural network for pulmonary disease prediction.

Area of Science:

• Artificial Intelligence
• Medical Imaging
• Computer Science

Background:

• Deep learning models are increasingly vital in healthcare for disease prediction.
• Traditional Convolutional Neural Networks (CNNs) can overlook dominant features in medical images.
• Accurate pulmonary disease diagnosis from X-rays is crucial for timely treatment.

Purpose of the Study:

• To develop an advanced deep learning model for precise pulmonary disease identification.
• To enhance feature extraction from chest X-ray images for improved diagnostic accuracy.
• To classify specific pulmonary conditions including fibrosis, pneumonia, and cardiomegaly.

Main Methods:

• A novel approach combining CNN with auction-based optimization algorithm (ABOA) for feature extraction.
• Utilizing the DSC process for classifying pulmonary disease types from X-ray images.
• Training and validating the model on the NIH Chest X-ray and ChestX-ray8 datasets.

Main Results:

• The proposed CNN-ABOA-DSC model demonstrated superior feature extraction capabilities compared to traditional CNNs.
• The approach achieved higher accuracy in predicting pulmonary diseases (fibrosis, pneumonia, cardiomegaly, normal) than existing state-of-the-art methods.
• Performance was validated against established deep learning techniques like BPD, DL, CSS-DL, and Grad-CAM.

Conclusions:

• The integrated CNN-ABOA-DSC approach offers a more accurate method for pulmonary disease prediction from X-rays.
• This novel technique effectively addresses limitations in feature extraction, leading to improved diagnostic outcomes.
• The findings suggest a significant advancement in AI-driven medical diagnostics for respiratory conditions.