A network fine-tuning step based on signal physics is proposed for deep learning based quantitative susceptibility mapping using high-pass filtered phase only to susceptibility. The proposed method showed better robustness compared to the pre-trained networks without fine-tuning when the test dataset deviated from the training dataset, such as a change in voxel size or high-pass filter cutoff frequency.
This abstract and the presentation materials are available to members only; a login is required.