xiat0616/brainageing

This is the project about simulating the ageing brain.

Learning to synthesise the ageing brain without longitudinal data

The proposed method synthesises the ageing version of a given brain with preserving subject identity. The project is the code of our MICCAI 2019 work and its extension in Medical Image Analysis.

Example results

Here we first show a result image. We generate the synthetically aged images given input image and target ages. Check our Journal paper for more details.

A video is also provided in the example results folder.

brainageing's Issues

Question about the Implementation of Wasserstein Loss

Dear Dr. Tian Xia,

Thank you so much for this wonderful work on brain ageing. I'm certain this will be of ongoing value to the research community as a whole. I am also working on predicting future brain MRIs, so reading your code is really helpful to me.

I have a question about the Wasserstein loss in your code. Why it is K.mean(y_true * y_pred)? I thought it should be loss = discriminator(fake) - discriminator(real) according to the WGAN paper.

I would very much appreciate it if you could point me in the right direction.

Thanks.

Pre-trained models

Would it be possible to share a pre-trained model of the generator?

Input dimensions & how to start training?

Hi! May I know what are the dimensions of the input image / condition vector (AD, MCI, CN) and age vector should be? Also, where should I start to run the training? Thanks!

Can I get requirements.txt file?