Deep Network Interpolation for Accelerated Parallel MR Image Reconstruction
This work addresses image quality trade-offs in medical imaging reconstruction, but it is incremental as it builds on existing unrolled and adversarial training methods.
The paper tackles the problem of accelerated parallel MR image reconstruction by introducing a deep network interpolation strategy, which models a trade-off between perceptual quality and fidelity by interpolating between models trained with different losses.
We present a deep network interpolation strategy for accelerated parallel MR image reconstruction. In particular, we examine the network interpolation in parameter space between a source model that is formulated in an unrolled scheme with L1 and SSIM losses and its counterpart that is trained with an adversarial loss. We show that by interpolating between the two different models of the same network structure, the new interpolated network can model a trade-off between perceptual quality and fidelity.