Abstract:

Medical imaging techniques such as magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), mammography, ultrasound and x-ray have been used for the diagnosis and treatment of diseases for many years. However, machine learning methods are used in this field for reasons such as earlier diagnosis of diseases, reduce the workload of doctors, and adjudicate conflicting expert opinions. With the increase in the amount of data, machine learning methods have remained insufficient in the field of image processing. Thanks to the developing mathematical models and hardware devices, deep learning has taken a wide place in this field. In this study, the application of deep learning methods in the field of medical image processing has been examined. Very recent examples are presented from studies in the fields of segmentation, classification and disease diagnosis, image generation, image transformation and image enhancement. The algorithms used in the studies are briefly explained. In addition, brain tumor segmentation with deep learning was tried on the BraTS 2020 dataset, and as a result, a dice similarity rate of 86% and a sensitivity value of 80% were obtained. Our aim is for this study to guide different studies on medical images with deep learning methods and serve as a basic resource in this field.

Keywords: