In more recent years, fatty liver disease (FLD) has been among the most common diseases around the world. However, FLD is linked to high mortality rates, timely diagnosis of the disease can help prevent the mortality rates and save lives. Several different methods have been proposed to diagnose the disease but still an improvement is needed in the results. Deep learning-based methods have provided promising results in medical sector. Therefore, in this work, we deployed AlexNet convolutional neural network for this purpose. This neural network has the ability to extract the optimal features of images due to convolutional layers with different kernels in the architecture. Also, in this thesis, transfer learning method was applied to increase the accuracy of diagnosing fatty liver disease. In the proposed approach, we transferred the extracted features to a multi-layer perceptron neural network optimized with the whale optimization algorithm. Finally, we were able to achieve 95% accuracy, which outperforms other state of the art methods in this field.
