Utilizing patient-specific 3D printed guides for graft reconstruction in thoracoabdominal aortic repair

In thoracoabdominal aortic aneurysm repair, repairing the visceral and segmental arteries is challenging. Although there is a pre-hand-sewn and multi-branched graft based on the conventional image-based technique, it has shortcomings in precisely positioning and directing the visceral and segmental arteries. Here, we introduce two new reconstruction techniques using patient-specific 3D-printed graft reconstruction guides: (1) model-based technique that presents the projected aortic graft, visualizing the main aortic body and its major branches and (2) guide-based technique in which the branching vessels in the visualization model are replaced by marking points identifiable by tactile sense. We demonstrate the effectiveness by evaluating conventional and new techniques based on accuracy, marking time requirement, reproducibility, and results of survey to surgeons on the perceived efficiency and efficacy. The graft reconstruction guides cover the segmentation, design, fabrication, post-processing, and clinical application of open surgical repair of thoracoabdominal aneurysm, and proved to be efficient for accurately reconstructing customized grafts.

www.nature.com/scientificreports/ available four-branched grafts considering the anatomical relationship based on the preoperative computed tomography (CT) images because the anatomical location of these vessels significantly vary among patients [15][16][17][18] . However, this conventional image-based approach has several shortcomings with respect to the precise positioning of the branching grafts because the accuracy depends on the constructor, usually, the operating surgeon. To overcome this limitation, 3DP-based reconstruction of the thoracoabdominal graft has been introduced to enable more accurate construction of the aorta and branching vessels 4 . As an extension of this approach, we developed two reconstruction techniques using patient-specific 3D printed graft reconstruction guides: (1) model-based technique (MBT), which uses a visualization model designed as a realistic-shaped graft model that contains the main aortic body and its branching vessels, making it possible to manually position the branching grafts on the artificial aortic graft and (2) guide-based technique (GBT) using the marking guides, wherein the branching vessels in the visualization model are substituted with slightly protruded marking points. We evaluated the accuracy and marking time efficiency of three techniques (conventional IBT, MBT, and GBT) in terms of the proper positioning of the aortic branching vessels, using a designed graft model (DGM) as the gold standard.

Methods
This retrospective study was carried out in accordance with the principles of Declaration of Helsinki and current scientific guidelines. The institutional review board for human investigations at Asan Medical Center approved this study with a waiver of informed consent from patients because of the use of retrospective clinical and imaging data. The data were de-identified, in accordance with the Health Insurance Portability and Accountability Act privacy rule. All methods were performed in accordance with the relevant guidelines and regulations.
The procedure for two types of graft guide techniques, MBT and GBT, are illustrated in Fig. 1. The visualization model and marking guides segmented based on CT images were designed, modelled, and exported into the standard tessellation language (STL) format. Then, they were printed using ColorJet printing (CJP) and the stereolithography apparatus (SLA) technology, following which they underwent post-processing and ethylene oxide gas sterilization. Graft reconstruction was performed using the MBT and GBT for clinical application and the accuracy and time requirements were evaluated.  Conventional image-based technique. The procedure of the conventional IBT proposed by Park et al. 14 is as follows: (1) In the axial view of the CT images, the position of the visceral and segmental (intercostal and lumbar) arteries are identified. (2) To evaluate the angle between the visceral and segmental arteries, a horizontal line and a virtual diverging line of the branch from the central spot of aorta are drawn on each axial slices of a CT image at the visceral and segmental levels, and the angles between both lines are measured at each level.
(3) Based on the visceral artery, the height of the segmental artery is determined and the corresponding point is marked with measured angle on the main aortic graft. (4) These processes are repeated as many times as the number of segmental arteries to be reconstructed.

Development of visualization model and marking guide.
In manually segmenting the native aorta and spine, the major aortic branching vessels-the celiac artery, superior mesenteric artery, renal artery, bilateral common iliac artery, segmental arteries-using Mimics Medical 17 and 3-matic 9.0 software (Materialise Inc, Belgium) ( Fig. 2a) were included. After 3D rendering based on the segmentation, the centreline of the native aorta was extracted through 3D computer-aided design (CAD) modelling and the squiggly centreline was modified to reflect the spine shape (Fig. 2b). The virtual model of the aorta was designed in the diameter of www.nature.com/scientificreports/ the prospective aortic graft to be implanted (24-30 mm); each of the virtual side-branches were united in 8 mm diameter as a single model. To achieve more efficient and accurate construction of the aortic graft, the graft reconstruction model and guide, visualization model, and marking guide, were developed: (1) The visualization model was designed as a realistic-shaped graft model containing the main aortic body and its branching vessels, to enable manual positioning of the branching grafts on the artificial aortic graft (Fig. 2c-left); (2) In the marking guide, the branching vessels in the visualization model were replaced by slightly protruding marking points (2 mm), to ensure that their positions were designed such as to be identifiable by tactile sense when wrapped by the aortic graft, instead of depending solely on visual mimicking ( Fig. 2c-right). The DGM was validated by operating surgeons, following which it was exported as STL files.
3D printing technology. The visualization model was fabricated using a ColorJet printer (CJP) with VisiJet PXL Core powder, VisiJet PXL clear binder, and colour bonding materials ( Fig. 2d-left). The marking guide was printed using a stereolithography apparatus (SLA) with Dental LT resin, a material capable of temporary mucosal contact ( Fig. 2d-right). To fabricate the graft reconstruction guides, the layer thickness, field of view and X-Y plane resolution for the CJP were 0.10, 254 × 381 × 203 and 300 × 450 DPI, respectively, and 0.10, 145 × 145 × 145, 0.15 mm, respectively, for the SLA. The X-Y resolution of SLA was not precisely defined; therefore, Formlabs www.nature.com/scientificreports/ Inc. experimented the specimens to find the X-Y resolution 19 . In addition, the spot size and power of laser were 140 μm and 250 mW.
Graft reconstruction method using MBT and GBT. In a real operative setting, the grafts were presewn by the operating surgeon with the patient during anaesthetic induction using either the MBT or GBT, and were used to surgically replace the diseased thoracoabdominal aorta (Fig. 2e). The postoperative configuration of the replaced aorta was confirmed by CT images (Fig. 2f). The experimental processes for patient-specific graft reconstruction using MBT and GBT are shown in Fig. 3. Under the supervision of the visualization model in the MBT, the shape of the main graft body was imitated (Fig. 3a) and the locations of the segmental vessels were indicated on the main graft body, with the point of the celiac artery origin as reference (Fig. 3b). In the GBT, the marking guide was inserted into the aortic graft and aligned with reference to the origin of the visceral arteries (Fig. 3c). Then, the positions of the segmental arteries were determined by the tactile sense through the protruding markers and marked on the graft surface (Fig. 3d). Both in the MBT and GBT, the segmental grafts were attached to the marked points by running sutures (Fig. 3e).

Survey.
To assess the convenience availed by the proposed techniques, three practicing cardiothoracic surgeons with experience in graft reconstruction with three techniques, IBT, MBT and GBT, were surveyed. The survey questionnaire was constructed in-house by curating aids to anatomical understanding, usefulness, satisfaction, surgical outcome, and recommendability from clinical standpoints. The surveyees had 3-10 years of experience as certified cardiothoracic surgeons. The questionnaire items were categorised as follows: (1) the level of contribution to understanding the anatomic structure (four questions); (2) usefulness for graft reconstruction (three questions); (3) satisfaction level for graft reconstruction (two questions); and (4) benefits of the use of 3DP technologies to surgical outcomes (two questions); and (5) recommendability of 3DP for thoracoabdominal aorta surgery to other surgeons (two questions) (Supplementary Table S1). The scores were recorded as 1 = Strong disagreement, 2 = Disagreement, 3 = Neutral, 4 = Agreement, and 5 = Strong agreement.
Evaluation and statistical analysis. The experiments of conventional IBT, MBT, and GBT performed by three researchers were measured, analyzed, and compared with the DGM based on the graft reconstruction accuracy and time required for fenestrations of vessels, excluding cutting out. This retrospective study was performed by applying three techniques to 15 patients, and mimic aortic grafts were fabricated to carry out the experiment. In the accuracy evaluation, the positions of the segmental arteries were indicated on the graft using the IBT, MBT and GBT (Fig. 4a), and the graft was unfolded using scissors (Fig. 4b). Then, the diagonal line, height, and angle between the designated celiac artery and the marked segmental arteries were measured (Fig. 4c). The Bland-Altman analysis was used to compare the differences in the diagonal line, height, and angles of the DGM and three techniques, IBT, MBT and GBT, and to identify the LoA, bias, and outliers 20,21 . The inter-group differences in the measurements and marking time were evaluated using the paired t-test. The correlation coefficients (r) for the DGM and three techniques were observed, and based on the correlation analyses of the same sample size by an independent group, it was determined that the two correlations were statistically significant. An intraclass correlation coefficients (ICC) analysis was performed on the IBT, MBT and GBT experiments by three researchers to confirm the correlations among them. In addition, the survey results on the IBT and the MBT or GBT were analysed using the Wilcoxon signed rank test. All the statistical analysis was conducted using statistics software (SPSS 25.00, IBM; and Medcalc 19.6.4 free trial, Medcalc Software).

Results
Comparison of accuracy and marking time requirement of DGM and three techniques. Table 2 presents the measurement of the diagonal line, height, and angle from the celiac artery to the segmental arteries for the DGM, whose design was validated by surgeons, conventional IBT, MBT, and GBT. According to the Bland-Altman analysis, the arithmetic mean and standard deviation (SD) of the difference between the DGM and conventional IBT were − 9.19 ± 16.47 mm (limits of agreement (LoA): − 41.47 to 23.08 mm) for the diagonal line, − 9.25 ± 16.84 mm (LoA: − 42.26 to 23.76 mm) for the height, and − 9.76° ± 10.05° (LoA: − 29.47° to 9.95°) for the angle (Fig. 5a-c) (Fig. 5d-f), respectively. The arithmetic difference values for the diagonal line, height, and angle between the DGM and GBT were − 1.00 ± 7.27 mm (LoA: − 15.25 to 13.24 mm), 1.79 ± 4.93 mm (LoA: − 7.86 to 11.45 mm), and − 6.39° ± 10.29° (LoA: − 26.55° to 13.78°) (Fig. 5g-i), respectively. There were statistically significant differences in all the measurements for the IBT, MBT, and GBT, except for the diagonal line of the GBT. The average marking time required for fenestrations of vessels excluding cutting out with three techniques are summarised in Table 2 Table 4 and Supplementary Table S1. The categories in the questionnaire were designed to rate graft reconstruction guides with the conventional IBT, MBT, and

Discussion
We developed two reconstruction techniques using patient-specific 3D printing graft reconstruction guides: (1) MBT using the visualization model designed based on the realistic-shaped graft model that contains the main aortic body and its branching vessels, making it possible to manually position the branching grafts on the  www.nature.com/scientificreports/ artificial aortic graft and (2) GBT using the marking guides, in which the branching vessels in the visualization model were replaced by slightly protruding marking points. The patient-specific 3D printed graft reconstruction guides may provide following benefits: (1) the marking time can be minimized by checking and marking the ideal position of the patient's segmental arteries in complex anatomy such as extremely tortuous spine and aorta.
(2) With the improved accuracy and reproducibility of the aortic graft reconstruction, the techniques can be utilized easily even for those in the learning curves. (3) Owing to the improved procedural efficiency, the techniques may contribute to reducing the level of fatigue of operating surgeons during lengthy and exhausting surgery, thus increasing focus on important procedures. Crawford extent II or III thoracoabdominal aortic aneurysm repair entails replacing the thoracoabdominal aorta and revascularising the visceral branches and intercostal or lumbar arteries simultaneously; therefore, it is regarded as the most extensive and challenging operation with high risks of surgery-related mortality and serious   22 . However, it is reportedly common for patch aneurysm to develop in the remnant aortic island, especially among those with connective tissue disorders such as Marfan syndrome [23][24][25][26][27] . Although there are limited published reports on employing the pre-sewn multi-branched aortic graft in open repair of thoracoabdominal aortic aneurysm, its use is now a desirable option to prevent such aneurysm formation as observed in the conventional island-type aortic repair 25,[27][28][29] . In addition, in cases where the ostia of the aortic branching arteries are displaced far away from each other, the use of a multi-branched graft may be the only way to effectively remove the diseased aorta. Furthermore, paraplegia is a devastating complication of thoracoabdominal aortic aneurysm repair, particularly in extent II and III, and preventing it is a primary focus. Segmental arteries between T8 and L2 have been considered to play an important role in spinal cord protection and majority of institutes routinely incorporate large intercostal arteries into the surgery to restore spinal cord perfusion 13,30 . However, those arteries are mostly reimplanted as a single patch through side-by-side anastomosis, even when pre-sewn thoracoabdominal aortic aneurysm graft is used, which also increases the risk of future patch aneurysm, particularly in Marfan patients. To reconstruct a thoracoabdominal aorta using multi-branched aortic graft personalized to reflect individual anatomical variation, we introduce the use of MBT of 3DP for the open surgical repair of thoracoabdominal aortic aneurysm with reasonable operative outcomes. Furthermore, recently, we incorporated the GBT into the MBT to complement it and improve the procedural efficiency and accuracy. The aim of this retrospective study is to quantitatively evaluate these techniques.
In addition to the complexity of this high-risk open repair, it is difficult to ensure that vital branch arteries are in the proper positions when constructing patient-specific grafts. To overcome these challenges, Park et al. developed the octopod technique, an eight-branched aortic graft for reconstruction based on the IBT 14 . However, because the graft was reconstructed by measuring the diagonal line, height, and angle based on CT images under anaesthetic induction in the operating room, it is difficult to understand the individual characteristics of patients in detail within a limited time, which affects the accuracy. For example, the second patient, who had Crawford extent II thoracoabdominal aortic aneurysm with severe scoliosis, had an extremely tortuous main aorta, affecting the visceral and segmental arteries, unlike the general anatomical structure 4 . Because the shape of the graft to be replaced should be connected to the aorta along the tortuous spine, the graft marked using the conventional IBT, calculated by slice thickness and number, may differ from the actual visceral arteries and segmental arteries positions. However, the MBT and GBT using graft reconstruction guides consisting of visualization model and marking guide efficiently reduce the graft marking time and solve anatomical hurdles. In our previous study, Rhee et al. 31 mentions that the experience on 3DP guidance in open surgical repair of thoracoabdominal aortic dissection and aneurysm have yet been limited in a small number of cases to allow adequate comparisons for its time efficiency in the real clinical practice. Nevertheless, of the 20 patients, 9 patients with 3DP had a shorter duration of cardiopulmonary bypass time than 11 patients, undergoing surgery without 3DP (median, 170 min vs 181 min) despite they had a greater number of segmental arteries revascularized (median, 2 vs. 3), which is mostly likely to require longer cardiopulmonary bypass times. Although the sample size is small, 3DP are assumed that the it is likely to give chance to more complete procedures with greater number of segmental artery anastomoses to prevent paraplegia and to decrease cardiopulmonary bypass time.
In this study, using three techniques, 135 mimic grafts (15 patients, three researchers, and three techniques) ranging from aortic arch to iliac artery were evaluated in experiments. The grafts are a very realistic mimic of standard cylindrical model grafts they were used to bypass the expensive graft cost. The DGM, conventional IBT, MBT, and GBT were evaluated and compared based on graft marking accuracy and time requirements. The errors of the experiments can be divided into marking error with medical images, 3DP errors, and human errors 10,32,33 . The marking errors of the conventional IBT were affected by the subjective measurement in CT images. The direction of the segmental arteries was measured from the centre points of the aorta and the segmental arteries, with the horizontal line. However, for the aneurysm patients, it manifested as elliptical, rather than circular, in the axial view, and the errors could be attributable to the incorrectness of the centre point of the aorta and the subjectivity of the centre point of the aorta and blood vessel ( Supplementary Fig. S3) 32,33 . The differences of MBT and GBT arose from 3DP technologies errors, including machine and 3D printed model and guide errors. The machine error is triggered by environmental factors, such as temperature, humidity, and vibration, and the materials, resolution, and usage duration of the 3D printers. The 3D printed model and guide errors depend on the size, shape, printing direction, and printing angle of the guides and post-processing, including support www.nature.com/scientificreports/ removal, ultraviolet polymerization, and surface smoothing 10,32 . In addition, autoclave and steam sterilization with high temperatures can deform or distort the graft reconstruction guide 34 . To minimize such problems, we opted for ethylene oxide gas sterilization at a relatively low temperature. Three techniques were prone to human errors; errors also occurred when defined landmarks-diagonal line, height, and angle-are measured using a digital calliper and digital angle ruler in the experiments by three researchers. Even if only two landmarks were selected among the diagonal line, height, and angle, the third could be derived; however, the landmarks were measured directly without using a mathematical formula considering measurement error. In the Bland-Altman analysis, the diagonal line, height, and angles in the conventional IBT at − 45.96 to 33.83 mm, − 46.70 to 33.34 mm, and − 39.05° to 13.05°, respectively, exhibited wider ranges of differences, compared to the MBT and GBT; the differences among the diagonal line and height tended to fall in the negative or positive region as the distance between the segmental artery and celiac artery increased, and the differences between the angles tended to be mostly distributed in the negative region from zero. Outliers tended to appear in the IBT as the distances from the celiac artery from the segmental arteries increased in the diagonal line and height, and occurred in patient with tortuous and swelling aneurysm aorta or scoliosis in the angle (Fig. 5a-c) Table 2). The diagonal line and height in the MBT and GBT were distributed in the region of the negative difference, as the measured distance was smaller, indicating a positive difference as the measured distance increased. The bias of the diagonal line and angles was greater in the GBT and the height was greater in the MBT (Fig. 5d-i). In the outliers of the MBT and GBT, the impact on the morphology of the aorta was greater than the distance of the segmental arteries. The MBT and GBT reduced the graft marking time significantly, and the effect was proportional to the number of segmental arteries ( Supplementary Fig. S4). For example, the graft marking time for the first patient with the least number of segmental arteries was 10.5, 0.5, and 2.8 min with the conventional IBT, MBT, and GBT, respectively. For the fourteenth patient with 13 segmental arteries, graft reconstruction time was 36.5, 1.4, and 4.4 min with the IBT, MBT, and GBT, respectively. The average time requirement for the MBT and GBT reduced by 17.4 and 15.5 min, respectively, showing the effect reduced by more than six times compared to IBT. In addition, although, based on the marking time, the MBT was more efficient than the GBT, the GBT had superior accuracy (Table 2 and Supplementary Table S2).
The correlation coefficient (r 14 ) for the diagonal line and height between the DGM and GBT was observed to be stronger than that of the IBT (r 12 ) and MBT (r 13 ), whereas a similar level of correlation was confirmed for the angles. Thus, it was implied that the GBT was the most similar to DGM (Table 3). In addition, the ICC was examined, to confirm the human error and the reproducibility of graft reconstruction among three researchers in conventional IBT and new techniques. All the techniques showed high agreement, with 0.995 for the GBT and 0.986 for the IBT and MBT.
For the conventional IBT (without 3DP), and the MBT and GBT (with 3DP), questionnaires were administered to three surgeons with experience in three techniques. Considering the survey score, MBT and GBT with the graft reconstruction guides could provide significantly better help to surgeons to the conventional IBT.
There are several limitations to this study. First, the procedures, from data acquisition to 3DP, lasted over a week, which made it difficult to apply the reconstruction techniques using patient-specific 3D printed graft reconstruction guides to open repair in patients with acute aortic rupture. Seconds, depending on the quality of the CT images, low-resolution images or slice thickness exceeding 3 mm made it difficult to identify the segmental arteries. Finally, improvements in efficacy and efficiency in the real clinical setting are the final purposes of using 3DP in the open repair of TAAA, however, the best available data on this issue have been limited by a small sized sample to allow adequate comparisons of hard clinical endpoints such as mortality, surgical bleeding, neurologic injuries and end-organ failures depending on the use of 3DP. In further studies, to reduce the graft reconstruction time, not only should the position of the segmental arteries on the graft be marked, functions, including the suturing and bonding process and the application for direct bioprinting process, should be developed. In addition, comparisons in clinical outcomes through larger cohort as the next step research are required to validate the usefulness of 3DP in open TAAA repair.

Conclusion
In this study, two types of grafts were developed for patient-specific graft reconstruction. Both methods provided a useful individualised approach to the anatomically challenging open repair. The MBT and GBT effectively reduced the graft reconstruction time and improved the accuracy, compared to the IBT.