A Study on Medical Image Analysis with its application in Disease Diagnosis
Loading...
Date
item.page.authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Medical Image processing generally deals with the images captured clinically to analyze its structural/geometrical properties for diagnosing diseases.In the process of disease diagnosis from raw images that are captured using different technologies like X-ray,Magneic Resonance Imaging,Computed tomography Scan(CT scan) is a challenging job due to the inherent noise that is present.The subsequent effect of dealing with the raw images without proper filtering is the wrong interpretation of it by an experienced practitioner.Therefore the preprocessing steps as denoisng of the raw images and segmentation of the Region of Interest are important.In the present thesis,we focus on diagnosis of human diseases from medical images along with sufficient preprocessing and analysis on it to provide a significant efficiency in the process.
newlineWe find after a thorough literature review on medical image processing and disease diagnosis, a number of challenging issues that still needs to be focused on and advanced research works.The limitations of the state of the art methods are mostly due to the complexity of medical imaging,the diversity of diseases and the ethical concerns is AI driven diagnosis.Hence some of the research gaps are mentioned below:
newlinei Development of an efficient diagnostic tool for rare diseases with minimal available training data. In this thesis we try to develop such diagnostic tools for dealing with skin cancer(melanoma) and Glaucoma for eye diseases.
newline
newlinei The current study uses transfer learning to classify diseases for improving the diagnostic capability in case of rare diseases.
newline
newlinePresence of unwanted noise that hamper the accuracy in diagnosis.In this study we discuss various techniques of preprocessing(denoising and segmentation) which helps in further image processing steps like feature extraction and classification.We provide an AI driven approach to perform image denoising. We carry out our study on how deep residual neural networks are able to denoise images such as salt noise, salt and pepper noise etc which ai