A SYSTEM AND METHOD FOR COVID- 19 DIAGNOSIS USING DETECTION RESULTS FROM CHEST X- RAY IMAGES

The present invention generally relates to a system and method for COVID- 19 diagnosis using detection results from chest X- Ray images. The system comprises a pre-processing unit for removing irrelevant information and noise from chest X- Ray images using a contrast limited adaptive histogram equalization (CLAHE) and extensive augmentation techniques; adconvolutional single shot detector (DSSD) in which SSD +Residual-101 is augmented with deconvolution layers for detecting small objects in the image; a feature extraction unit configured for extracting features for generating maps representing spatial, semantic information, and right transformations; and a convolutional neural network (CNN) technique for segmentation and classification of chest X- Ray (C: XR) images available openly, and thereafter attempted to detect if a C: XR image corresponds to a COVID-19 infection. Pre-processingUnit Deconvolutional 102 SingleShot Detector Feature Extraction Convolutional Neural Unit 106 Network 108 Figure removing irrelevant information and noise from chestX- Ray images using a contrast lmitedada ptivehistogramequalization[CLAHE)a dextensiveaugmentaton 202 r techniques employing a deconvolutional single shot detector (DSSD) in which SSD +Resdual-101 204 is augmentedwithdeconvolution layers fordetecting small objects in the image extractingfeaturesforgeneratingmapsrepresentirigspatial,semanticinformation, 206 and right transformations segmenting andclassifying chestX- Ray (C: XR) Images using a convolutional neural 208 network(CNN)technique,andthereafterattemptedtodetect fa C:XR image corresponds toa COVID-19 infection Figure b c d Figure 3