Style Transfer using Generative Adversarial Networks (GAN)
Project description-
To ensure a better diagnosis of patients, doctors may need to look at multiple MRI scans. What if only one type of MRI needs to be done and others can be auto-generated?
- Different MRIs are required for different abnormalities. A single type of MRI may not be sufficient for the diagnosis of an abnormality. Additional MRIs might enhance the accuracy of diagnosis, leading to better treatment of the patient. but Access to different imaging techniques is difficult and expensive. Moreover, doctors may advise getting one type of MRI to be done at a time, which makes it a time-consuming process. with the help of an exciting tool in the deep learning domain, which is known as Generative Adversarial Networks or GANs.
- Generative Adversarial Networks (GANs) have been used for generating deepfakes, new fashion styles and high-resolution pictures from the low-resolution ones.
- GANs can be used in the field of medical science, for instance, to create a different type of MRI from an existing one. a particular variant of GANs, called CycleGAN, is used to translate the style of one MRI scan into another, such as T1-weighted to T2-weighted or vice versa.