SNRAware: Improved Deep Learning MRI Denoising with Signal-to-Noise Ratio Unit Training and G-Factor Map Augmentation.

Purpose To develop and evaluate a deep learning-based MRI denoising method using quantitative noise distribution information obtained during image reconstruction to improve model performance and generalization. Materials and Methods This retrospective study included a training set of 2 885 236 images from 96 605 cardiac cine series acquired with 3-T MRI scanners from January 2018 to December 2020. Of these data, 95% were used for training, and 5% were used for validation. The hold-out test set included 3000 cine series, acquired in the same period. Fourteen model architectures were evaluated by instantiating each of the two backbone types with seven transformer and convolution block types. The proposed SNRAware training scheme leveraged MRI reconstruction knowledge to enhance denoising by simulating diverse synthetic datasets and providing quantitative noise distribution information. Internal testing measured performance using peak signal-to-noise ratio and structural similarity index measure, whereas external tests conducted with 1.5-T real-time cardiac cine, first-pass cardiac perfusion, brain, and spine MRI assessed generalization across various sequences, contrast agents, anatomies, and field strengths. Results SNRAware improved performance on internal tests conducted on a hold-out dataset of 3000 cine series. Models trained without reconstruction knowledge achieved the worst performance metrics. Improvement was architecture agnostic for both convolution and transformer models. However, transformer models outperformed their convolutional counterparts. Additionally, three-dimensional input tensors showed improved performance over two-dimensional images. The best-performing model from the internal testing generalized well to external samples, delivering 6.5 and 2.9 times contrast-to-noise ratio improvement for real-time cine and perfusion imaging, respectively. The model trained using only cardiac cine data generalized well to three-dimensional T1-weighted magnetization-prepared rapid gradient-echo brain and T2-weighted turbo spin-echo spine MRI acquisitions. Conclusion The SNRAware training scheme leveraged data obtained during the image reconstruction process for deep learning-based MRI denoising training, resulting in improved performance and good generalization.