Low PSNR (~17) on IXI Dataset for T1-to-T2 Translation

[dariusoneof]

I encountered an issue where the PSNR for T1-to-T2 translation on the IXI dataset is only around 17, which is significantly lower than the reported values.
I have followed the exact experimental parameters from the command-line example provided in the repository and have not modified the code.
Could you please provide some insights into what might be causing this discrepancy?

[emirahmed123]

I encountered an issue where the PSNR for T1-to-T2 translation on the IXI dataset is only around 17, which is significantly lower than the reported values. I have followed the exact experimental parameters from the command-line example provided in the repository and have not modified the code. Could you please provide some insights into what might be causing this discrepancy?

Are you experiencing offsets in your output images? We are using SynDiff to produce some synthetic medical images and we are noticing that there is a translation of 2-3 pixels in the outputs in comparison to the source images. This is resulting in low PSNR scores.

[AriaZiyanYang]

Can we have the preprocessing code?

[xnw5mm4xd4-spec]

I encountered the same issue, may I ask how you solved it?

[xnw5mm4xd4-spec]

I encountered an issue where the PSNR for T1-to-T2 translation on the IXI dataset is only around 17, which is significantly lower than the reported values. I have followed the exact experimental parameters from the command-line example provided in the repository and have not modified the code. Could you please provide some insights into what might be causing this discrepancy?

Are you experiencing offsets in your output images? We are using SynDiff to produce some synthetic medical images and we are noticing that there is a translation of 2-3 pixels in the outputs in comparison to the source images. This is resulting in low PSNR scores.

May I ask how you solved it?