Evaluation of accelerated motion-compensated 3d water/fat late gadolinium enhanced MR for atrial wall imaging

Figures

• 

  Fig. 1. Overview of the accelerated motion-compensated 3D iNAV-based water/fat LGE imaging approach. A 2D iNAVs are acquired before 3D dual-echo undersampled data acquisition and used to estimate superior-inferior (SI) and right-left (RL) translational respiratory motion of the heart by tracking a template located around the atria (red rectangle). B SI motion is used to group the 3D data in a set of respiratory bins, and SI and RL motion estimates are used to correct intra-bin translational motion. Images reconstructed at each respiratory position are then used to estimate non-rigid deformation fields. C The binned data and motion fields are used in a motion-corrected reconstruction framework and the resulting motion-corrected dual-echo 3D images are then used to compute the final water/fat 3D LGE images

• 

  Fig. 2. Visual comparison of 3D LGE images acquired with the conventional dNAV-based approach and the iNAV-based approach for a representative patient without LGE findings. iNAV-based 3D LGE reduces inflow artefacts in the pulmonary vein present in the dNAV images (red arrow). Acquisition times (TA) are expressed as min:s

• 

  Fig. 3. Visual comparison of 3D LGE images acquired with the conventional dNAV-based approach and the iNAV-based approach for a representative patient with observable atrial fibrosis (green arrows). Motion-induced artefacts of intensity similar to the LGE findings can be observed in the dNAV-based images (red arrow). Acquisition times (TA) are expressed as min:s

• 

  Fig. 4. Image quality scores for (a) dNAV-based and (b) iNAV-based images for three expert readers. Average score for dNAV-based images was 2.6 ± 1.0 while for iNAV-based images was 3.1 ± 1.0. For reader 1 there was no statistically significant difference between methods (p = 1), while for readers 2 and 3, iNAV images were significantly better (p = 0.0078 and p = 0.0088 respectively). (Image quality 1: Non-diagnostic, 2: Acceptable, 3: Good, 4 Full Diagnostic)

• 

  Fig. 5. Visual comparison of 3D LGE images acquired with the conventional dNAV approach and the iNAV-based approach for a representative patient. Motion-induced artefacts in the dNAV-based images prevented this image from being used for diagnosis. Acquisition times (TA) are expressed as min:s

• 

  Fig. 6. Proposed iNAV-based water/fat LGE atrial imaging protocol for three representative patients showing an example water LGE image, fat image, and a fused version of both (with fat depicted in red). The proposed imaging method provides a fully co-registered complementary fat image that can be used to quantify peri-atrial fat

• 

  Fig. 7. Atrial shells for two representative patients showing 3D visualization of fibrosis from dNAV and iNAV images, with fibrotic areas depicted in red. Example slices for each patient are also displayed, where fibrotic areas are indicated by red arrows. Similar scar patterns are seen between the dNAV and iNAV-based images; however, differences in contrast due to the difference in acquisition times post contrast administration result in some differences in the scan patterns observed in the rendered 3D images. RUP right upper pulmonary; LUP left upper pulmonary; RLP right lower pulmonary; LLP left lower pulmonary; LA left atrial