A 59-year-old female with the history of hypertension who was diagnosed with chronic TBAD with ARSA in 2015 was admitted. The ARSA arose from the proximal portion of the descending thoracic aorta. After 6 years of optimal medical treatment, she was readmitted and operated on because of 1-month intermittent dull chest pain. The primary entry was located near the ARSA. A re-entry tear was detected at the beginning of the left subclavian artery (LSA) (Figure 1E). The patient refused open surgery or hybrid repair. Therefore, EMSBSG, specifically designed for aortic arch lesion, was used to seal the tears and hence restore the aortic morphology. The EMSBSG system was manufactured by Hangzhou Endonom Medtech Co., Ltd (Hangzhou, China) according to individual needs. In the current study, the EMSBSG system is comprised of the main aortic stent graft (34 ^* 26 ^* 160 mm) with an embedded branch port for ARSA, the stented port for LSA, and the branched stent graft, as shown in Figure 1B. Detailed information on the steps to perform the endovascular repair procedure using EMSBSG system is presented in the Supplementary Data Sheet 1. The EMSBSG system is in the clinical trial, and full details of the clinical studies are available at ClinicalTrials.gov (NCT04765605). This study was approved by the Institutional Review Board of the Chinese PLA General Hospital (S2020-010-01). Written informed consent was provided by this patient prior to this study.

The patient underwent three CTA scans before the operation (Pre-1) and at 4 and 25 days after EMSBSG implantation (Post-1 and Post-2). All three CTA datasets were acquired via a dual-source CT scanner (Brilliance iCT256, Royal Philips, Dutch). Detailed scanning parameters are available in Supplementary Data Sheet 2. DICOM files of Pre-1, Post-1 and Post-2 CTA datasets were exported with slice numbers of 614, 1,219 and 641, respectively. Image segmentations and three-dimensional vessel configuration reconstructions were conducted via the automatic image processing module developed by our team, with the segmentation time of 14.19 ± 2.15 s for each case (9), as shown in Figure 1A. The image segmentations were subsequently reviewed for each slice by an expert with rich experience in reading CT images with supervision by one professional vascular surgeon. Manual correction was performed as necessary. The luminal region with patency was segmented, and the cross-sectional contours of reconstructed geometries were mapped back to CTA images to ensure that the 3D-reconstructed models presented the actual outline of the vessel lumen. For each patient, segmentations were performed from the ascending aorta to the distal end of the dissection. The reconstructed models were finally exported in triangulation mesh (STL format) to facilitate the calculation of any morphological parameters. Figure 1C shows the 3D-reconstructed aortic models of Pre-1, Post-1 and Post-2.

Morphological parameters were measured on the basis of 3D-reconstructed models. To quantify the geometric variations of different models, an optimal alignment of the Pre-1, Post-1 and Post-2 models is a prerequisite procedure. The iterative closest point algorithm, which is the most popular method for 3D rigid registration (10), was used to transform postoperative models to properly align with the preoperative model. The algorithm was developed via the Visualization Toolkit (VTK) package in Python 3.9 (https://vtk.org).

When the preoperative and postoperative models were aligned, luminal volume changes for the true lumen (TL) and false lumen (FL) were calculated (Figure 1D) and centerlines were also extracted. To further investigate the efficacy of EMSBSG insertion, the morphological changes of the main aortic grafting region (MAGR) were also computed including volume changes and the parameters along the centerline of MAGR. In detail, a series of slices perpendicular to the centerline with an interval of 1.0 mm were first extracted (Figure 1F) and then the morphological characteristics of each slice were computed including the area, circumference, equivalent diameter, transverse diameter, longitudinal diameter and the ratio between the transverse and longitudinal diameters (aspect ratio). The curvature and tortuosity of MAGR were also computed on the basis of the centerline. Tortuosity was defined as equation 1 (Eq. 1), where d indicates the linear distance of the centerline and l indicates the distance along the centerline. The methods used to measure each parameter are further described in Supplementary Data Sheet 3. The normal deformation index of MAGR from Post-1 to Post-2 was calculated to quantify luminal expansion over time.

Time-variant velocities at the ascending aorta, right common carotid artery, left common carotid artery, LSA and ARSA were measured via Doppler ultrasound of the patients. At each measurement site, an appropriate ultrasound probe was employed (Supplementary Table 1; Supplementary Data Sheet 4). The upper edge of the velocity sonogram was extracted as the variation in the maximum velocity at the measured site (Figure 1G). The measurements were used to provide patient-specific velocity boundary conditions for the computational model. In the current study, Doppler ultrasound data were acquired before treatment and at 4 days after treatment. Therefore, Post-1 and Post-2 were given the same boundary conditions. Supplementary Figure 2 shows the velocity boundary conditions for Pre-1, Post-1 and Post-2 (Supplementary Data Sheet 4). The flow distribution ratio for each aortic arch branch, including the right common carotid artery, left common carotid artery, LSA and ARSA was computed on the basis of the Doppler ultrasound data, which was defined as the ratio of the branching flow to the inflow. Details of pulsatile waveforms of pressure at the celiac artery, superior mesenteric artery, renal arteries and outlets at iliac arteries were obtained from a previous study and used as pressure boundaries (11). In our previous study (12), measured pressure was compared with the data we used in this study, confirming the rationality of pressure boundary conditions.

3D aortic models were meshed using ICEM (Ansys Inc., Canonsburg, USA) with tetrahedral elements in the core region and prismatic cells (10 layers) in the boundary layers near the aortic wall. The grid resolutions for Pre-1, Post-1 and Post-2 were 3397577, 2211420 and 2329161 cells, respectively. The values of the cardiac cycle extracted from Doppler ultrasound examination were 55 and 85 beats/min for pre- and post-treatment data, respectively. Temporal discretization of numerical models was assigned to be 50 steps per cycle. The transport equations of time-dependent flow were numerically solved using a finite volume solver, CFD-ACE (ESI Group, Paris, France) (Supplementary Data sheet 5). The blood was assumed as Newtonian and incompressible with a density of 1,044 kg/m³ and dynamic viscosity of 0.00365 kg m⁻¹ s⁻¹. For simulation purposes, the no-slip and rigid arterial wall was ascertained. For each model, four cardiac cycle simulations were carried out to obtain a periodic solution, and the results of the final cycle were presented for postprocessing and analysis. Grid and time step independence tests were conducted in our previous study (13, 14), confirming the rationality of the base grids and base temporal resolution in this study.

Qualitative and quantitative analyses were used to investigate the hemodynamic patterns at Pre-1, Post-1 and Post-2. The aortic pressure distribution and flow patterns at the systolic peak were studied. Wall shear stress (WSS) related parameters were selected to analyze the impact of EMSBSG insertion, including time-averaged wall shear stress (TAWSS), oscillatory shear index (OSI) and relative residence time (RRT). Details of the computation of WSS-related parameters are presented in Supplementary Data Sheet 6.

The luminal pressure difference (LPD) and the first balance position (FBP) were computed, which have been proven to be an efficient functional indicator of TBAD in our previous study (15). The spatial variation of LPD along the aorta and shift of the FBP between Pre-1, Post-1 and Post-2 were assessed. The total energy loss was evaluated in this study. The energy difference between the inlet and outlets of the aorta indicated the total energy loss during a cardiac cycle (16), which was defined as follows.

Where Q is the blood flow rate. TP indicates the total pressure.

Where u and P refer to velocity and pressure, respectively.

The alterations in systolic peak pressure, TAWSS, OSI and RRT from Post-1 to Post-2 of MAGR were computed. In the current study, Δ was used to denote the difference of the abovementioned parameter between Post-1 and Post-2, which was defined as Eq. 4.

Where X is the previously mentioned parameters. Area ratios (AR) with positive value (AR_P) and negative value (AR_N) were subsequently computed for each ΔX, which was defined as Eq. 5.

To capture different aspects of the local hemodynamics, parameters including systolic peak pressure, TAWSS, OSI and RRT were averaged across the circumferences orthogonal to the centerline of MAGR and ARSA (every 1 mm).

Morphological characteristics before and after EMSBSG insertion were quantified. As shown in Figure 2A, both the volume of TL and MAGR increased, while the volume of FL decreased. A slight increase in tortuosity was detected from Pre-1 to Post-1 with values of 0.222 and 0.231, respectively. However, the tortuosity reduced to 0.211 in Post-2, which was smaller than that in Pre-1. Furthermore, Figure 2C indicates the fluctuations of the curvature. The results showed that the curvature of Pre-1 and Post-1 is not only larger than that of Post-2 but also changes more dramatically than that of Post-2. The area, circumference and equivalent diameter along the MAGR centerline showed an overall increase except a slight reduction in the embedded branch port at Post-1. The region of the embedded branch port is indicated between two dotted lines in Figure 2; while Post-2 showed luminal expansion compared with Post-1. The results of longitudinal diameter, transverse diameter and aspect ratio showed that the EMSBSG system could enlarge the vessel lumen. Both Post-1 and Post-2 showed an increase in transverse diameter and aspect ratio, and Post-2 was larger than Post-1, as shown in Figures 2H,I. The tear information for this patient at different time-points were concluded in Supplementary Table 2; Supplementary Data Sheet7.

Pressure distributions and velocity streamlines for Pre-1, Post-1 and Post-2 at the systolic peak are shown in Figure 3, where the color map of pressure and velocity magnitude were restricted to a certain range for all cases to assist visualization. Post-1 showed higher pressure than Pre-1. The pressure dropped at the distal region near the EMSBSG region from Post-1 to Post-2. However, the pressure increased in both TL and FL in the abdominal region. The ARSA grafting region showed a continuous pressure rise after treatment. At the systolic peak, fast and organized flow was found in TL; while vortical and relatively slow flow was present in FL. The flow accelerated in MAGR of Post-1, while delayed in Post-2. In the ARSA grafting region, the velocity showed a reduction from Pre-1 to Post-1, while high-speed flow was detected in the root of ARSA in Post-1. In Post-2, the flow has a significant slow-down.

Loading patterns are crucial for hemodynamic analysis of vessel remodeling after stent-graft implantation. Figure 4 shows the distributions of TAWSS, OSI and RRT of Pre-1, Post-1 and Post-2. Before treatment, we observed a high TAWSS concentration in the root of ARSA (Figure 4A). With EMSBSG insertion, TAWSS elevated in the both MAGR and ARSA grafting regions at Post-1 while declined at Post-2, as shown in Figures 4B,C. OSI contour plots are shown in Figures 4D–F. There were some high OSI regions before surgery, while OSI decreased in the EMSBSG region at Post-1, and Post-2 showed a further reduction. Figure 4G displays the low RRT of the preoperative model, while high RRT appeared in the FL region of Post-1, as indicated in Figure 4H. The zone showed high RRT in Post-1 thrombosed in Post-2, as shown in Figure 4I.

To investigate the LPD between the TL and FL, a series of slices perpendicular to the centerline of the TL were extracted and the net pressures in the TL (P_TL) and FL (P_FL) on each slice over a cardiac cycle were calculated. Figure 5A displays the LPD (LPD = P_TL – P_FL) for each case. The FBP of LPD shifted to the abdominal aorta distally in Post-1, with a shift distance of 20.172 cm from Pre-1 to Post-1. Moreover, the FBP of the LPD shifted out of the dissected region at Post-2. The energy loss was compared among Pre-1, Post-1 and Post-2 (Figure 5B). Pre-1 showed the lowest energy loss with a value of 5.19 W and it increased to 8.39 W after EMSBSG procedure. However, a slight drop was observed in Post-2 at a value of 7.86 W. The flow distribution ratio showed a general increase for each aortic arch branch after EMSBSG intervention, as shown in Figure 5C.

Examinations of the spatial differences in peak systolic pressure, TAWSS, OSI and RRT demonstrated the patterns along the MAGR and ARSA regions (Figure 6). In MAGR, pressure and TAWSS increased after EMSBSG placement, while declined slightly at Post-2 (Figures 6A,B). Both OSI and RRT showed a significant decrease from Pre-1 to Post-1 (Figure 6C). In the ARSA grafting region, the peak systolic pressure showed a continuous rise after EMSBSG implantation, as shown in Figure 6E. TAWSS and RRT increased first at Post-1 and then declined at Post-2 (Figures 6F,H). The hemodynamic parameter alterations from Post-1 to Post-2 were further analyzed (detailed later).

To quantify the prognostic improvement, the evolution fields obtained by registration are shown in Figure 7. Figure 7A displays the normal deformation index contour plot, which illustrates the vessel lumen changes over time after EMSBSG placement. Figures 7B–E show the alterations of systolic peak pressure, TAWSS, OSI and RRT from Post-1 to Post-2, respectively. Figure 7F displays the AR of each previously mentioned parameter. The AR_N was larger than the AR_P for all factors, indicating that all of these parameters decreased from Post-1 to Post-2, which might designate a good prognostic result in MAGR.