Pneumonia detection on chest X-rays from Xception-based transfer learning and logistic regression.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Machine Learning

Background:

• Pneumonia is a leading cause of death globally, particularly in children and the elderly.
• Chest X-rays are a primary diagnostic tool, but interpretation by radiologists is time-consuming and prone to errors, especially with overlapping symptoms with other conditions like COVID-19.
• Current machine learning models for pneumonia detection from X-rays struggle with data imbalance and generalization, limiting their accuracy and robustness.

Purpose of the Study:

• To improve the accuracy and robustness of machine learning models for pneumonia detection from chest X-rays.
• To address data imbalance issues through data augmentation techniques.
• To evaluate the effectiveness of transfer learning for automatic feature extraction compared to traditional hand-crafted features.

Main Methods:

• Utilized transfer learning with pre-trained models (Xception and VGG-16) for automatic feature extraction from chest X-rays.
• Employed data augmentation to mitigate data imbalance problems.
• Trained and compared various classifiers, including Support Vector Machine, Logistic Regression, K-Nearest Neighbor, Stochastic Gradient Descent, Extra Tree Classifier, and Gradient Boosting Machine.

Main Results:

• Transfer learning-based features demonstrated superior performance compared to hand-crafted features for pneumonia detection.
• Achieved a high accuracy of 99.23% for pneumonia detection using chest X-rays with the proposed transfer learning approach.
• The study successfully addressed data imbalance issues, leading to more robust model performance.

Conclusions:

• Transfer learning offers a powerful approach for automatic feature extraction in medical image analysis, significantly enhancing pneumonia detection accuracy.
• Data augmentation is crucial for improving the performance of machine learning models dealing with imbalanced datasets in medical diagnostics.
• The developed method provides a highly accurate and efficient automated solution for pneumonia detection from chest X-rays, potentially aiding clinical decision-making.