Tuberculosis detection in chest X-ray using Mayfly-algorithm optimized dual-deep-learning features.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computer Science

Background:

• Tuberculosis (TB) is a leading cause of death globally, necessitating early and accurate diagnosis.
• Chest radiography (X-rays) is a primary tool for diagnosing pulmonary TB.
• Deep learning (DL) offers potential for automated analysis of medical images.

Purpose of the Study:

• To develop and validate an automated deep learning scheme for detecting Tuberculosis (TB) in chest X-ray images.
• To enhance classification accuracy for TB detection using advanced DL techniques.
• To compare the performance of different feature extraction and selection methods in TB diagnosis.

Main Methods:

• Image collection and pre-processing of chest radiographs.
• Feature extraction using pre-trained VGG16 and VGG19 models.
• Optimal feature selection employing the Mayfly Algorithm (MA).
• Serial feature concatenation to create dual-deep-features (DDF).
• Binary classification using a 5-fold cross-validation with a K Nearest-Neighbor (KNN) classifier.

Main Results:

• The proposed DL scheme was evaluated using VGG16, VGG19, conventional features, optimal features, and concatenated dual-deep-features (DDF).
• The DDF approach, combining features from VGG16 and VGG19, demonstrated superior performance.
• The system achieved a high classification accuracy of 97.8% for TB detection using the KNN classifier.

Conclusions:

• The automated deep learning system utilizing dual-deep-features (DDF) shows significant promise for accurate and efficient TB detection in chest X-rays.
• The Mayfly Algorithm effectively optimized feature selection, contributing to improved diagnostic accuracy.
• This approach can support clinicians in early TB diagnosis, potentially improving patient outcomes.