Development of Deep Convolutional Neural Network Algorithms Based on DenseNet Model for Adaptive Classification of Chest Diseases

Chest diseases pose serious health threats if not detected early for properly and timely diagnosis. These diseases could include pulmonary disease, pneumonia, asthma, tuberculosis, lung diseases and other cardiovascular complications. When patients exhibit symptoms like chest pain, shortness of breath, or persistent cough; physicians often prescribe chest X-rays to assess the underlying cause of their discomfort. Chest X-ray is a commonly used diagnostic imaging test that requires significant expertise and careful observation to classify the particular or multiple chest diseases due to the complex nature of the pathology and fine texture of lung lesions. Chest X-Ray classification is a challenging and time consuming task in medical image classification due to the complexity of the human chest structure and the subtle variations in X-Ray images caused by different medical conditions. This paper presents a novel classification technique for the classification of 14 chest diseases that could impair patient’s health. The classification technique presented in this paper is a deep convolutional neural network (CNN) algorithms based on DenseNet model structure for adaptive classification of chest diseases. This paper address and bridges three main research gaps in chest diseases classification, namely: 1). enhancing the classification accuracy and specificity in diagnosing up to 14 different classes of chest diseases based on label distributions; 2). develops an AI-based deep learning CNN using DenseNet model structure to accurately classify and predict the risk probability levels of heart failures based on chest X-ray images; and 3). the research leverages the issues of interpretability and explainability as the results is self-explanatory to clinicians. The results demonstrate the efficiency, proficiency and robustness of the deep CNN algorithm based on DenseNet model structure. The paper concludes with a comprehensive discussion on the model’s performance, shedding lights on its strengths and potential areas for improvement. The technique presented in this paper can easily be adapted for real-time chest diseases classification by clinicians.