CT imaging
Preoperative and postoperative evaluation of LA, LAA and adjacent structures was performed by CCTA (Somatom Definition, Siemens Medical Solutions, Forchheim, Germany). The temporal resolution is 330 ms, the detector collimation is 64*0.6 mm, the tube voltage of 120 kV and tube current of 380 mA. 100 ml contrast medium was injected with 50 ml saline flush followed through elbow vein at the rate of 5 ml/s.
Image analysis was performed by the Extended Brilliance Workspace version 4.5 (Philips Healthcare Cleveland, OH, USA). All images were analyzed by two blinded, experienced radiologists. Evaluating the consistency of the measurement results between the two radiologists, the mean values of each CT indicator were taken for follow-up analysis. In case of disagreement in the evaluation process, the two radiologists reevaluated and discussed to reach a consensus.
Measurements were made of the maximum diameter of LAA orifice at the preoperative CCTA images. The double-oblique orthogonal view of LAA orifice is obtained by the method of multi-planar imaging reconstruction (Figure 1). The orifice was defined as the plane between the circumflex artery and a point 15 ± 5 mm from the tip of the limbus, reflecting as closely as possible the site where the proximal aspect of the Watchman device would be expected to expansion.
Preoperative left atrial volume (LAV) and left atrial appendage volume (LAAV) were measured by Mimics Medical 17.0 (Materialise NV, Leuven, Belgium). The thinfilm cross sectional images generated by CCTA were imported into the Mimics in the DICOM data format. Images at the end systolic cardiac phase (when the LA cavity was largest) were selected as original data. Tissues connected to LA were then separated based on 3D model. Separation of LA and left ventricle was bounded by the mitral valve annulus. Pulmonary veins (PV) and LAA were separated by the PV ostia and LAA orifice, respectively (Figure 1). LAV and LAAV were automatically calculated by Mimics. The specific details were described as our previous essay[6].
PDL and device endothelialization were evaluated at the postoperative CCTA images. CT workspace was used to reconstruct the original CCTA images in 3D multi-planar, and the axial plane should be at the level of LA. For the Watchman device, first observe the position of the nitinol skeleton of the device. Afterwards, move the coronal axis within the transverse window perpendicular to the coves of the parachute of the device. Afterwards, align the axes on the two other viewers also perpendicular to the coves of the parachute of the device. Lastly, the center of the axes should be placed to the center of the screw-hub. With all of the above imaging steps, the LAA occluder view for post-implantation evaluation (LOVE) could be established[7].
Quantitative contrast assessment of LAA was carried out by measuring the average linear attenuation coefficient (Hounsfield units [HU]) in the LAA distal to the implanted device, using a circle diameter of 3 mm for the region of interest. PDL was defined as the average linear attenuation coefficient of LAA was more than 100 Hu, and continuous contrast enhancement was observed from LA to LAA along the side of the device[8]. Incomplete endothelialization was defined as the average linear attenuation coefficient of LAA was more than 100 Hu, and continuous contrast enhancement was observed from LA to LAA through the fabric of the device (“fabric leak” from diffusion of contrast through the non-endothelialised polyethylene terephthalate membrane). The average linear attenuation coefficient less than 100 Hu was defined as LAA complete occlusion[7].