Automatic Chest X Ray Pathology Detection using Convolutional Neural Network

This research study presents an automated Chest X-ray Pathology Detection system employing Convolutional Neural Networks (CNNs) and transfer learning. Here, the primary research objective is to develop a robust medical diagnosis system for chest Xray images, capable of distinguishing among 14 distinct pathologies commonly found in these images, including lung infections, tumors, fractures, and cardiovascular conditions. This study utilizes the Keras framework and a pre-trained DenseNet CNN model for enabling efficient and accurate classification. This study involves comprehensive data preparation, transfer learning with DenseNet, addressing class imbalance, and model evaluation. The proposed system demonstrates promising results and potential for enabling significant improvements in medical diagnostics.