3-Dimensional Gradient and Spin-Echo Magnetic Resonance Cholangiopancreatography with Deep Learning Reconstruction at 3 T: Achieving Superior Image Quality with Reduced Acquisition Time

Purpose To evaluate image quality and clinical feasibility of breath-hold 3D magnetic resonance cholangiopancreatography (MRCP) using a gradient and spin-echo (GRASE) technique with deep learning reconstruction (GRASE-DLR) versus GRASE without DLR and turbo spin-echo with DLR (TSE-DLR) at 3 T. Methods Sixty-five consecutive patients who underwent 3D MRCP on a 3-T system were retrospectively enrolled. Three protocols were compared: GRASE-DLR, GRASE without DLR, and TSE-DLR. Acquisition time, quantitative metrics (SNR and CNR), and five-point qualitative scores for overall image quality, artefact reduction, background suppression, and ductal visualization were independently assessed by two radiologists. Interobserver agreement was evaluated using Cohen's weighted kappa. Diagnostic performance for biliary and pancreatic disease and anatomical variations was evaluated. Subgroup analysis was performed for patients with poor breath-hold capacity (n = 8). Results Mean acquisition time for GRASE-DLR was 8.9 s, representing reductions of 45.4% versus TSE-DLR and 49.1% versus GRASE (both p < 0.001). Despite comparable SNR across all three protocols, GRASE-DLR achieved significantly superior overall image quality, artefact reduction, and major duct visualization (all p < 0.001), with moderate-to-substantial interobserver agreement. Sensitivity for biliary disease was markedly higher with GRASE-DLR (93.8%) than TSE-DLR (56.2–62.5%) and GRASE (68.8–75.0%), with accuracy of 89.5–94.7%. For pancreatic disease, sensitivity was 87.5% with GRASE-DLR versus 55.0–75.0% for comparators, with accuracy of 84.8–87.0%. Indeterminate biliary anatomical variation findings were nearly eliminated with GRASE-DLR (0–1.5% vs. 21.5–26.2% for TSE-DLR). In patients with poor breath-hold capacity, GRASE-DLR demonstrated pronounced improvements in image quality and artefact suppression. Conclusions GRASE-DLR achieves superior image quality with substantially reduced acquisition time and improved diagnostic confidence, particularly in patients with limited breath-hold tolerance.