Adenoid segmentation in X-ray images using U-Net

Use of machine learning and specifically deep learning-based techniques for medical diagnosis has created a significant impact on early and easier diagnosis in the domain of radiology. Deep learning techniques have demonstrated unprecedented superiority in all facets of medical image analysis ranging from classification to identification to segmentation. The efficacy of deep learning algorithms to process X-ray data and extract meaningful information from it has helped diagnose and provide timely health care to patients. The focus of proposed work is to use DICOM x-ray images for detection and segmentation of adenoid gland using deep learning-based techniques. The distance between the Adenoid gland and soft palate may be used by doctors to identify the severity and type of diseases and hence an automated method for identification of adenoid gland will help in automatic diagnostics. The main challenge is that the size and shape of adenoid gland varies with age and disease and hence because of its deformative property it is difficult to segment it. In this work, we propose to use U-net based technique for segmentation of adenoid gland. To the best of our knowledge this is the first attempt to solve the problem of adenoid detection and segmentation using U-net based deep learning architecture.