Can Virtual Contrast Enhancement in Brain MRI Replace Gadolinium?: A Feasibility Study

Area of Science:

• Radiology
• Artificial Intelligence
• Medical Imaging

Background:

• Gadolinium-based contrast agents (GBCAs) are widely used in MRI but their use is increasingly scrutinized.
• There is a need for contrast-free MRI techniques to reduce GBCA administration.
• Deep learning (DL) offers potential for predicting contrast enhancement from noncontrast scans.

Purpose of the Study:

• To investigate the feasibility of predicting contrast enhancement from noncontrast multiparametric brain MRI scans.
• To develop and evaluate a deep-learning (DL) architecture for virtual contrast enhancement prediction.
• To assess the potential of reducing GBCA application through AI-driven contrast enhancement prediction.

Main Methods:

• A Bayesian DL architecture was developed using 10-channel multiparametric MRI data.
• The model predicted virtual contrast enhancement and was evaluated against ground truth contrast-enhanced T1-weighted imaging.
• 116 datasets from glioma patients and healthy subjects were used, with tumor subregions analyzed.

Main Results:

• The virtual contrast enhancement prediction achieved 91.8% sensitivity and 91.2% specificity.
• T2-weighted and diffusion-weighted imaging sequences were most influential for prediction.
• Quantitative metrics (AUC, PSNR, SSIM) demonstrated strong performance, though qualitative evaluation showed some visual differences compared to real contrast enhancement.

Conclusions:

• The developed model for virtual gadolinium enhancement shows strong quantitative and qualitative performance.
• Further studies in larger, diverse patient groups are needed to confirm its clinical utility in reducing GBCA exposure.
• AI-driven virtual contrast enhancement holds promise for optimizing GBCA use in clinical practice.