Assessment of the generalization of learned image reconstruction and the potential for transfer learning

FIGURE S1 Overview of the training data that were used in this study. A, Data from a coronal PD_w 2D turbo spin-echo sequence for knee imaging with and without FS. These two data sets show substantial differences in terms of image contrast and SNR. B, Knee data from the PD_w sequence without FS, with additional noise added to the complex multichannel k-space data, such that the SNR is comparable to the lower SNR FS data. Completely synthetic k-space data were generated from images from the BSDS

FIGURE S2 Zoomed regions of interest from the results shown in Figure 2, further demonstrating the noise amplification and blurring in cases of SNR mismatch between training and test data and the effects of joint training with heterogeneous data. The selected ROI highlights these effects in an anatomical region that includes complex image texture due to bone trabeculae and fine details due to ligaments and cartilage.

TABLE S1 Quantitative analysis of SSIM and RMSE to the fully sampled reference for the generalization experiments with respect to contrast and SNR. The mean and SDs are shown for all cases in the test set, consisting of 378 slices for PD_w and PD_w with additional noise. The PD_w FS test set consisted of 365 slices. Highest image quality values are achieved when data from the same sequence are used for both training and testing. Drops can be observed when the SNR level deviates between training and testing. Results for trainings with PD_w FS data and PD_w data with added noise are comparable. Joint trainings from multiple contrasts and SNR levels show comparable performance to using consistent data between training and testing.

TABLE S2 Quantitative analysis of SSIM and RMSE to the fully sampled reference for the experiments with different numbers of training examples for joint training. The mean and SDs are shown for all cases in the test set, consisting of 378 slices for PD_w and 365 slices for PD_w FS.

TABLE S3 Quantitative analysis of SSIM and RMSE to the fully sampled reference for the generalization experiments with respect to the sampling pattern. The mean and SDs are shown for all cases in the test set, consisting of 378 slices for PD_w and 365 slices for PD_w FS. Trainings show good generalization ability with respect to changes in the sampling pattern. In particular, for non-FS test cases, the SSIM values are identical for all combinations of sampling patterns in training and testing.

TABLE S4 Quantitative analysis of SSIM and RMSE to the fully sampled reference for the trainings with synthetic data. The mean and SDs are shown for all cases in the test set, consisting of 378 slices for PD_w and 365 slices for PD_w FS. A slight drop in image-quality values can be observed in comparison to trainings with in vivo data. The experiments show the same behavior in terms of SNR dependency.

TABLE S5 Quantitative analysis of SSIM and RMSE to the fully sampled reference for the experiments with different numbers of training examples for the transfer learning experiments. The mean and SDs are shown for all cases in the test set, consisting of 378 slices for PD_w and 365 slices for PD_w FS. Transfer learning fine-tuned results outperform both baseline trainings using only synthetic data, as well as reference trainings using the same subset of knee MRI data that were used during fine-tuning. Quantitative image quality values are close to trainings with consistent in vivo data.