AI-powered Gradient Echo Plural Contrast Imaging (AI-GEPCI): a Comprehensive Multiparametric Neurological Protocol from a Single MRI Scan
Authors
Affiliations (1)
Affiliations (1)
- Washington University School of Medicine
Abstract
BackgroundMRI plays an essential role in diagnosing and monitoring neurological diseases. Conventional protocols rely on multiple sequences to obtain complementary contrasts, increasing scan time, cost, and tolerability. Generating multiple contrasts from a single acquisition may streamline workflow while maintaining clinical utility. PurposeTrain attention-based convolutional neural networks (ACNNs) to generate clinical-quality FLAIR, MPRAGE, R2*, and derived contrasts from a single Gradient Echo Plural Contrast Imaging (GEPCI) acquisition, enabling multi-contrast imaging from one scan. Study TypeRetrospective. Population43 MRI scans from individuals with multiple sclerosis (25/18 F/M, 49{+/-}11 years old). Field Strength/Sequence3T MRI was used to obtain 3D GEPCI, MPRAGE, and FLAIR sequences. AssessmentTechnical quality of the AI-generated contrasts was evaluated against directly acquired MRI images using structural similarity index (SSIM). Quantitative accuracy for R2* maps was evaluated using normalized root-mean-square error (NRMSE). Clinical image quality was assessed by expert physicians. Lesion volumes and counts were obtained using automated segmentation. ResultsAI-generated FLAIR and MPRAGE images achieved mean SSIM values of 0.923{+/-}0.028 and 0.935{+/-}0.022, respectively. The generated R2* maps achieved a mean SSIM of 0.996{+/-}0.006, with quantitative accuracy reflected by an NRMSE of 0.031{+/-}0.020. Physicians rated GEPCI-FLAIR images at 4.2 and GEPCI-MPRAGE images at 4.5 (on a 1-to-5 scale), both exceeding the clinically routine standard of 4.0. Lesion volume and count comparisons from automated segmentation showed strong agreement between AI-generated and ground-truth measurements (R{superscript 2}=0.988 and R{superscript 2}=0.933, respectively). ConclusionAI-GEPCI generated multiple clinically relevant MRI contrasts from a single GEPCI acquisition with high similarity to corresponding acquired images. Radiological reviews and quantitative analyses supported the feasibility of producing high-quality, intrinsically co-registered multi-contrasts for comprehensive brain evaluation.