Over the past years, lung diseases have received particular attention, especially after the COVID-19 pandemic, which entailed more research on such a vital organ. This made promising solutions arise using computer vision to try diagnosing abnormalities. This approach focused on optimizing Deep Learning (DL) models to obtain more detailed and better results classifying 14 different chest diseases, such as Pneumonia, Nodules, and Hernia. We used a DenseNet model with 121 layers plus various data augmentation techniques and Adamax optimizer and learning rate decay trained on NIH chest X-ray dataset containing over 112,000 frontal view images. Results were slightly better than others’ approaches as far as we know in some classes and a better overall AUROC score exceeding the state-of-the-art score.