Augmented data

Case contributed by Dr Candace Makeda Moore

Augmented data from brain MRI

MRI

The images are almost indistinguishable from normal, unremarkable T1 weighted MRI axial images from a brain.

Case Discussion

The images were created from a single axial image in a T1 weighted MRI for the purposes of data for machine learning. They are examples of augmented data. In this specific case, the images were created via a computer algorithm written in Python which samples an original image and creates new augmented ones by simple matrix operations. Another type of augmented data, not shown, can be created by adding noise to the original data. 

Although the images look indistinguishable from normal radiological images to the human eye, a well designed computer algorithms might be able to detect that they were made from re-sampling original image data and multiplying it by a specific matrices. 

How to use cases

You can use Radiopaedia cases in a variety of ways to help you learn and teach.

Creating your own cases is easy.

Updating… Please wait.

 Unable to process the form. Check for errors and try again.

 Thank you for updating your details.