LUNG OPACITY IDENTIFICATION USING MATHEMATICAL MODEL BASED ON DEEP LEARNING

25 Abstract In this paper, “Identifying Lung Opacity Using a Mathematical Model Based on Deep Learning,” I presented a deep learning architecture for diagnosing lung disease during chest X-rays. I used the recently created RSNA lung disease dataset. This is the ImageNet dataset. This dataset is the first and so far the only dataset to specialize in the detection of constitutive areas of Lung Opacity on chest x-ray and will allow us to train neural networks specialized in classification and object detection tasks. I used the Convolutional Neural Networks (CNN) algorithm because CNN-based deep learning classification approaches have the ability to automatically extract highlevel representations from big data using little preprocessing compared to other image classification algorithms. The Xception model was used for the classification tasks. At the same time, this project introduced learning strategies and prediction strategies using the Python and Keras deep learning library.