Radiological Image Synthesis Using Cycle-Consistent Generative Adversarial Network

EasyChair Preprint 5338

Abstract

Radiology is the branch of science that deals with the study of energetic radiations and their use in generating medical images. MRI(Magnetic Resonance imaging) and CT (Computed tomography) are the two widely used modalities in radiology. CT comes with the disadvantage of high radiation risk which may have side effects. Thus, medical image from MRI only radiation which is much safer than CT can be used to synthesize CT images using Deep Learning techniques. In this paper we, propose to build an architecture of Fully Convolutional neural network (FCN) along with a cyclic Generative Adversarial network(GAN). Our model has successfully generated CT images from the given MRI images from an unpaired ADNI image dataset.

Keyphrases: ADNI dataset, CGAN, CT generation, FCN, MRI, Radiology, deep learning

@booklet{EasyChair:5338,
  author    = {Rahul Nehra and Abhisikta Pal and B Baranidharan},
  title     = {Radiological Image Synthesis Using Cycle-Consistent Generative Adversarial Network},
  howpublished = {EasyChair Preprint 5338},
  year      = {EasyChair, 2021}}