Currently the script here for the Siemens reconstruction pipeline does straightforward zero-padding and FFT to account for partial Fourier undersampling. By comparing with the DICOM images that the scanner creates, it looks like the Siemens default approach is to apply some kind of k-space filter (probably a ramp filter) before applying the FFT. This reduces ringing and increases apparent SNR at the cost of a decrease in apparent spatial resolution. Please comment here if you have any ideas for what an improved default behaviour would be for the retroMoCoBox.
Currently the script here for the Siemens reconstruction pipeline does straightforward zero-padding and FFT to account for partial Fourier undersampling. By comparing with the DICOM images that the scanner creates, it looks like the Siemens default approach is to apply some kind of k-space filter (probably a ramp filter) before applying the FFT. This reduces ringing and increases apparent SNR at the cost of a decrease in apparent spatial resolution. Please comment here if you have any ideas for what an improved default behaviour would be for the retroMoCoBox.