Image Reconstruction in Surgical Field Using Deep Learning

Abstract

The field of medical image reconstruction helps to improve image quality by manipulating image features and artefact with Filtered-Back Propagation for X-ray Computer Tomography (CT), Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET). This project focuses on detection of tumour cells using Radiomics application that aims to extract extensive quantitative features from magnetic resonance images. In this paper image discretization models and image interpolation techniques are used to segment the MR images and train them for Image Reconstruction. The image based gray level segmentation is carried out for required feature extraction to improve the clustering analysis for segmentation. Convolution Neural Network is used for image classification and recognition because of its high accuracy. The CNN follows a hierarchical model which works on building a network and finally gives out a fully-connected layer where all the neurons are connected to each other and the output is processed. The JPEG approach is a commonly used type of compression of lossy images that centres on the Discrete Cosine Transform. By splitting images into components of varying frequencies, the DCT functions. Finally the output from the Radiomics application is compared with the existing methodology for determining the Mean Squared Error - Loss Function to ensure the image compression quality.