Detection of Pneumonia in Chest X-ray using Deep Learning Enabled Adaptive and Altruistic PSO-based Feature Extraction

Children die from pneumonia in many underprivileged areas around the world. The diagnosis of pneumonia in developing countries is still a difficult task, even with simple instruments available and medications available to treat it. Pneumonia can be identified in these nations using computer-aided diagnostic (CAD) technologies. We apply the ideas of deep learning and meta-heuristic algorithms to our pneumonia detection CAD system. In the first stage, we extracted the deep features by using the ResNet models. We performed feature extraction using various ResNet modelsResNet50, ResNet101, and ResNet152-to identify which model is better at extracting the features, and it can then be fine-tuned based on a target pneumonia dataset. After performing the feature selection procedure by particle swarm optimization, we modify the parameter using involvement of memory in adaptation and enrich it by developing agents that exhibit altruistic behavior We also consider adaptive and altruistic particle swarm optimization (AAPSO) for efficient feature selection and remove the non-informative features extracted by the ResNet model to improve the ability to detect pneumonia. In addition to ResNet models for feature extraction and AAPSO for feature selection, we applied the Support Vector Machine (SVM) algorithm for classification, as SVM is widely used for better classification and regression purposes.