Deep Learning Techniques to Analyze Chest X-Ray and Providing Accurate Predictions Regarding the Presence of Lung Disease

Medical research is starting to leverage the availability of radiological images for attempting to predict lung diseases with the use of images. The goal of the prediction is to identify pneumonia, tuberculosis, or lung cancer on CT scans and chest Xrays. For classification, Convolutional Neural Networks are considered the leading method in image processing; hence they will be utilized to automatically classify the images. To tackle the challenges associated with restricted datasets and class imbalance, more advanced approaches such as data augmentation and transfer learning will be used as well. The achieved accuracy and robustness metrics suggest that the system is useful for aiding practitioners with prognosis and diagnosis. This system can also assist in enhancing the accuracy of clinical decision making and minimizing the burden placed on the vise by implants on primary lung image analysis. Future work will consider combining other types of data like the patient's records or DNA to further improve accuracy.