モーション補正付きサブスペース再構成とナビゲーターを用いた3テスラにおける定量的高分解能スパイラル初回通過心筋灌流イメージング

To optimize a spiral acquisition using a fixed-angle subspace navigator and golden angle trajectory, combined with motion-corrected (MOCO) subspace reconstruction, for motion-robust, quantitative high-resolution whole-heart first-pass perfusion imaging at 3 T.

Spiral acquisition used a fixed-angle navigator to extract the temporal basis, combined with seven spiral arms ((1 + 7)-arm) rotated using k-t golden angles, yielding a 90 ms temporal footprint for two interleaved slices at 1.3 × 1.3 mm². Non-rigid deformation fields estimated from auxiliary images were incorporated into the subspace reconstruction. Performance was tested in XCAT simulations and 5 retrospectively undersampled free-breathing datasets. Prospective reconstructions using Subspace, L1-SENSE, and SENSE with and without MOCO from 22 patients were visually graded (1-5) by three cardiovascular imagers. In 11 of these patients an arterial input function (AIF) was acquired and myocardial blood flow (MBF) maps were calculated using Fermi-function deconvolution.

In simulations and retrospective datasets, the navigator captured cardiac motion and contrast dynamics effectively, and the (1 + 7)-arm k-t GA DD1-Hs configuration improved subspace reconstruction through increased spatiotemporal incoherency. In prospective studies, Subspace-MOCO achieved the highest visual scores (p < 0.001), with MOCO improving all methods (EMM ± 1.96 × SE: SENSE-MOCO 2.2 ± 0.2, L1-SENSE-MOCO 4.1 ± 0.2, Subspace-MOCO 4.6 ± 0.12). Subspace-MOCO also provided superior boundary sharpness and global edge sharpness, higher NCC and NMI (p < 0.05). Its MBF values were similar to those obtained with SENSE-MOCO and closer than those reconstructed from L1-SENSE-MOCO.

The proposed navigator-guided MOCO subspace reconstruction substantially reduces respiratory motion artifacts and enables high-resolution, motion-corrected whole-heart quantitative spiral perfusion imaging.