pneumonia Detection Using Deep Learning

Pneumonia is a severe form of lung infection which can easily become fatal if misdiagnosed or diagnosed late for an efficient treatment.Old diagnostic systems used to rely on X-ray images of the chest and the decision time which were excessively long and almost exclusively human-dependent that in turn made them susceptible to various biases.This work/effort presents a way of constructing the pneumonia detecting system through deep learning and its core is the promotion of both the accuracy and the efficiency of input phase of diagnostics, such as the diagnosis process.In order to provide a comprehensive evaluation, three deep learning models, VGG19, MobileNetV2, and a custom CNN model, will be described and performance comparison will be made using the chest X-ray dataset.Every choice of architecture will maximize their specific advantages and implement the collective strengths of the models to achieve better results.The suggested model will receive an original X-ray image, prepare it for model processing, and then report the confidence scores after the prediction has been made.However, based on test and validation data sets, the ensemble methods seem to be more efficient and accurate than the individual models as these give a better fit to the correctness of the data and surpass the performance of all individual models being applied to the validation set.The new approach is to build a Flaskbased web app that will incorporate these models and allow users to upload an X-ray image and how the models have decided about the image as well.This work can also explore possible future research, for example, including additional modalities to the dataset, and even carry out clinical examinations to affirm the results of the research.