Generative Adversarial network (GAN) for Virtual contrast enhancement in MRI

(University Hospital Erlangen, Germany)

Hemorrhage segmentation using U-Net

Brain hemorrhage should be treated as a medical emergency.      
What if the medical center doesn’t have an expert to look into CT images..?

Here, I trained an U-net model to segment brain hemorrhage from CT images.
The data as taken from Kaggle and annotations are done by me (with my limited knowledge).

Even though the data set contains only 200 samples, 
an image augmentation technique and transfer learning worked well..

B-OPSY Lung cancer detection from human breath

(Bachelors Project: Sahrdaya College Of Engineering and Technology, Kerala India )