Generating Virtual Post-Contrast T1 and ECV Maps from Native T1 in Cardiac MRI

An AI model generated virtual post-contrast T1 and extracellular volume (ECV) maps from native T1 maps without gadolinium, achieving a 3.05% mean RMSE for myocardial ECV versus reference, though with wide limits of agreement (-5.98% to +6.06%).

• A self-consistent recursive diffusion bridge framework was developed and tested on 813 cardiac MRI slices from patients with suspected myocardial disease.
• Virtual ECV maps showed acceptable agreement with reference ECV in the held-out test set (R² = 0.585), but with notable slice-to-slice variability.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.