Purpose: While MRI contrast is predetermined by imaging protocol, it can be retrospectively tuned, which is potentially useful for normalizing MRI data for radiomics. In this study, we present a new paradigm to obtain various contrasts from a single T1-weighted image.
Methods: We propose a contrast tuning framework that combines deep learning-based quantitative MRI with Bloch equations. Using deep neural networks, quantitative tissue relaxation parametric maps (T1 map, proton density map) and field map (B1 map) are predicted from a single T1 weighted image. Here, ground truth T1 map was obtained from four variable-flip-angle T1-weighted images and a B1 map (measured using the actual flip angle method); proton density map was calculated from T1-weighted image and T1 map. In these tasks, self-attention convolutional neural network is employed with unique shortcuts equipped and attention mechanism integrated. A total of 1,344 knee images from 56 subjects are utilized for training and testing in six-fold cross validation. Given estimated parametric maps, MR images with different contrasts are generated with the application of Bloch equations. While a wide spectrum of contrasts can be obtained with various imaging parameter values, the result is only validated at certain contrasts (corresponding to variable flip angles).
Results: High accuracy has been achieved in quantitative parametric maps and qualitative images. From a T1-weighted image, quantitative T1 map, proton density map and B1 map are predicted with an averaged correlation coefficient of 0.95~0.99 and L1 error of 0.02~0.09; and T1-weighted images corresponding to alternative flip angles are obtained with an averaged correlation coefficient of 0.97~0.99 and L1 error of 0.04~0.09.
Conclusion: A new data-driven strategy is proposed for retrospective MRI contrast tuning from a single T1-weighted image, which requires no additional data acquisition.
Not Applicable / None Entered.