Synthetic multi-inversion time magnetic resonance images for visualization of subcortical structures

Area of Science:

• Neuroscience
• Medical Imaging
• Artificial Intelligence

Background:

• Subcortical gray matter visualization is crucial for neuroscience research and clinical applications like disease diagnosis and surgical planning.
• Multi-inversion time (multi-TI) T1-weighted (T1-w) MRI significantly improves visualization but is not widely available in public datasets or common clinical settings.
• Existing MR datasets often lack the specific multi-TI sequences needed for detailed subcortical analysis.

Purpose of the Study:

• To develop a deep learning method, SyMTIC (synthetic multi-TI contrasts), capable of generating synthetic multi-TI MR images from routinely acquired sequences.
• To enable enhanced visualization and analysis of subcortical gray matter using widely available MRI data.
• To overcome the limitations of multi-TI sequence availability in public datasets and clinical practice.

Main Methods:

• SyMTIC utilizes deep neural networks for image translation, combining standard T1-w, T2-weighted (T2-w), and fluid-attenuated inversion recovery (FLAIR) images.
• The method estimates longitudinal relaxation time (T1) and proton density (ρ) maps, leveraging imaging physics principles.
• These estimated maps are then used to synthesize multi-TI images with user-defined inversion times.

Main Results:

• SyMTIC accurately synthesized multi-TI images from standard clinical inputs, achieving image quality comparable to explicitly acquired multi-TI data.
• The generated synthetic images, particularly for inversion times between 400-800 ms, significantly enhanced the visualization of subcortical structures.
• Segmentation of thalamic nuclei was improved using the synthetic multi-TI images.

Conclusions:

• SyMTIC provides a robust method for generating high-quality multi-TI MR images from routine clinical contrasts.
• The approach demonstrates generalization across varied clinical datasets, even those lacking FLAIR or T2-w images, and unknown parameters when paired with the HACA3 algorithm.
• SyMTIC offers a practical solution for improving brain MR image visualization and analysis in neuroscience and clinical settings.