Certain investigations on deep learning variants for infectious disease detection and diagnosis in medical images

Abstract

Pandemic diseases such as COVID-19 have significantly increased newlinethe demand for computerized medical image analysis for swift and accurate newlineresults. The infected regions are scanned through diagnosis devices that newlinegenerate X-ray, Computed Tomography (CT), Magnetic Resonance Imaging newline(MRI), or Positron Emission Tomography (PET) images. The images are newlineinduced for analysis using conventional digital image processing techniques newlinewith optimization algorithms. The advantages of such methods are the newlineprecision diagnosis through internal and external correlations, training, and newlinetesting and validation measures. Medical image analysis pursues the default newlineprocessing steps such as pre-processing, feature extraction, segmentation, newlinedetection, etc. In the case of pandemic disease detection from input images, newlinethe differences in regions are identified between regions concerning the pixel newlinedistribution. In such a process the infected regions show up changes in newlinetextural patterns that are identified using heterogeneous features. The change newlinein pixel distribution, feature arrangement, and dimensions are optimal in newlinedeciding the infection level, progress, and diagnosis. This research proposal newlineintroduces three novel methods for infectious disease diagnosis using CT newlineimages each of which performs its kind. newline