Synthetic Multi-inversion Time Magnetic Resonance Images from Routine Clinical Contrasts

Savannah P Hays, Lianrui Zuo, Anqi Feng, Yihao Liu, Blake E. Dewey, Jiachen Zhuo, Ellen M Mowry, Scott D Newsome, Jerry L Prince, Aaron Carass

Keywords: MRI, image synthesis, brain

TL;DR: A model for multi-inversion time MR image synthesis using routinely acquired contrasts.

Registration Requirement: Yes

Abstract:

Visualization of subcortical gray matter is essential in neuroscience and clinical practice, particularly for disease understanding and surgical planning.
Multi-inversion time (multi-TI) T$_1$-weighted (T$_1$-w) magnetic resonance (MR) imaging improves visualization of these structures, but is not available in common public datasets.
We present SyMTIC (Synthetic Multi-TI Contrasts), a physics-informed deep learning framework that generates multi-TI images from routinely acquired T$_1$-w, T$_2$-weighted (T$_2$-w), and FLAIR images.
SyMTIC estimates quantitative longitudinal relaxation time ($T1$) and proton density ($\rho$) maps, which are used within an inversion recovery signal model to synthesize MR images at arbitrary inversion times.
This formulation enables flexible contrast generation beyond fixed targets such as FGATIR.
On an in-domain dataset ($N=23$), SyMTIC produces high-quality synthetic images, achieving a PSNR/SSIM of $45.70 \pm 5.67$ / $0.9970 \pm 0.0029$ for MPRAGE and $27.60 \pm 2.27$ / $0.7906 \pm 0.0535$ for FGATIR.
The synthesized contrasts improve visualization of subcortical structures and support downstream tasks such as thalamic segmentation.
Additionally, by incorporating HACA3-based harmonization and imputation, SyMTIC generalizes to out-of-domain datasets and scenarios with missing input contrasts.
These results demonstrate that SyMTIC provides a practical and flexible solution for enhancing MR image contrast and analysis using standard clinical acquisitions.

Reproducibility: https://github.com/shays15/symtic

Visa & Travel: No

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Submission Number: 67