Deep-Learning Accelerated Vessel Wall Imaging Using T1-SPACE at Ultra-High Field Strength MRI.
Authors
Affiliations (1)
Affiliations (1)
- From the Department of Radiology, Mayo Clinic, Rochester, MN, USA (G.B., P.R., Y.E.Y.C, A.J.F, J.C.B, J.H., S.A.M), Research and Clinical Translation, Magnetic Resonance, Siemens Healthineers AG, Erlangen, Germany (M.D.N), and Siemens Medical Solutions USA, Rochester, MN, USA (P.K., K.D).
Abstract
To evaluate the feasibility and technical performance of a post-contrast T1-SPACE sequence using deep-learning based image reconstruction (DLBIR) for intracranial vessel wall imaging (IC-VWI) at 7T, comparing image quality, artifacts, and acquisition time with the standard-of-care (SOC) T1-SPACE sequence. In this retrospective single-center study, 36 patients (21 women; mean age, 53.3 ± 16.2 years) underwent IC-VWI at 7T using both SOC T1-SPACE and DLBIR-accelerated T1-SPACE sequences. Two independent neuroradiologists assessed overall image quality (noise, artifacts, sharpness, and overall quality), wall and lumen visualization along the intracranial vessels using a 4-point Likert scale. A cumulative-logit mixed-effects model (CLMM) of segment ratings was used for intersequence comparison. Segments were also pooled into proximal vs distal vessels and quality compared using exact paired sign test. Overall image quality metrics were compared using paired Wilcoxon tests and Bland-Altman plots. Inter-reader agreement was summarized using percent agreement. On segment-level analysis, T1-SPACE<sub>DL</sub> yielded markedly higher ratings than T1-SPACE<sub>SOC</sub> [wall: OR = 22.79 (95% CI 15.83- 32.82), lumen: OR = 97.0 (95% CI 66.8-141.0); both FDR-adjusted p<0.001]. Effects remained large when segments were pooled into proximal and distal segments [For wall, proximal segment OR = 297.0 (95% CI 40-∞) and distal segment OR = 71.0 (28-404); for lumen, proximal OR = 82.0 (36-339) and distal OR = 225.0 (60-13350); all FDR-adjusted p<0.001]. Reader-wise Wilcoxon tests showed higher overall image quality scores across noise, artifacts, sharpness, and overall quality for T1-SPACE<sub>DL</sub> ([all p < 0.001]). Inter-reader agreement was uniformly high with T1-SPACE<sub>DL</sub> when compared to T1-SPACE<sub>SOC</sub>. Bland-Altman analysis demonstrated a positive bias favoring T1-SPACE<sub>DL</sub>. Acquisition time was reduced from 7:30 minutes with T1-SPACE<sub>SOC</sub> to 6:00 minutes (20% reduction) with T1-SPACE<sub>DL</sub>. Contrast-enhanced T1-SPACE<sub>DL</sub> at 7T may have potential clinical utility for IC-VWI, given the shorter acquisition, significantly improved image quality and reduced image artifacts. DLBIR＝deep learning-based image reconstruction; SNR＝signal-to-noise ratio; CNR＝contrast-to-noise ratio; VWI＝Vessel Wall Imaging; DL＝ deep-learning; SOC＝ standard-of-care.