PDF
ePub
Citation
Print
Since its first introduction in clinical practice more than a decade ago, endovascular repair of abdominal aortic aneurysm (AAA) has been widely performed and it is reported to be an effective alternative to conventional open surgery, especially for the patients with comorbid medical conditions (1-3). Unlike open surgical repair of AAA, successful completion of endovascular repair largely depends on the medical imaging, and helical CT angiography (CTA) has been confirmed to be the preferred modality for both the preoperative planning and the postoperative follow-up of endovascular aortic repair (4, 5).
The development of multislice CT (MSCT) has provided important advantages over single slice CT with regard to CT angiography of the aorta and aortic stent grafting (6, 7). MSCT enables faster scans than single slice CT by providing high-volume coverage and thin-section images within a single breath hold, and this results in improved spatial resolution in the longitudinal plane. MSCT has been reported to be superior to single slice CT for nearly all clinical applications (6). Helical CTA for imaging the aorta and stent grafting has been complemented by a series of 3D postprocessing reconstructions, including the surface shaded display (SSD), maximum-intensity projection (MIP), curvilinear reformation (CVR), volume rendering (VR) and virtual intravascular endoscopy (VIE). These 3D reconstructions have been applied for both the pre-and postoperative assessment of aortic stent grafting. However, the 2D axial CT images still remain the standard reference in most of the pre-operative situations such as measurements of the aneurysm sac and neck diameter, visualization of aortic wall calcification and etc (8). In contrast, the 3D CT images were found to be useful for visualization of the relationship between the stent graft and the arterial branches (9-11). Assessing the diagnostic performance of these 3D reconstructions will assist clinicians to choose the appropriate image visualization for assessing endovascular aortic repair, and also to make efficient use of the MSCT imaging modality for clinical purposes. In this pictorial essay, we discussed the diagnostic value of each of the 3D reconstructions that were based on MSCT angiography scanning for a group of patients with AAAs that were treated with aortic stent grafting.
Patient Data and Multislice CT Scanning Protocol
The study was performed on 18 patients (15 men and 3 women, mean age: 75, age range: 63-84) who underwent endovascular repair of abdominal aortic aneurysms between 1999 and 2001. The patients were recommended to receive stent graft treatment from the vascular surgeons because they were unsuitable for open surgical repair due to comorbid medical conditions (hypertension and cardiovascular disease). All the patients were treated with the Zenith AAA Endovascular Graft (William Cook Europe, Bjaeverskov, Denmark) with a suprarenal uncovered component placed above the renal arteries for obtaining proximal fixation. The stent graft is an endoprosthesis constructed of nitinol monofilament that's woven into a tubular zigzag configuration. A thin woven polyester fabric was used to cover the nitinol wire frame. The patients were referred for follow-up examinations at one week, one month, three months, six months and one year after endovascular, and then annually. The initial CT scan was performed on a single slice CT scanner, and the most recent model that was used was on a multislice CT scanner. Therefore, only MSCT data were used for the 3D reconstructions. MSCT angiography was performed on a 16-slice scanner (Toshiba Medical Systems Europe, Netherlands) with the following scanning protocol: 16×0.75 mm beam collimation, a pitch of 2.0 and a reconstruction interval of 1 mm. MSCT angiography was performed with an intravenous injection of 100 ml of non-ionic contrast media (Niopam 300, Bracco UK Ltd. High Wycombe) administered at a rate of 2 ml/second with a scan delay of 30 seconds.
3D Image Generation
The CT volume data were converted from the original DICOM (Digital Imaging and Communication in Medicine) images with using commercially available software Analyze V 5.0 (www.Analyzedirect.com Mayo Clinic, USA). Bony structures were removed by using the region of interest drawings for generating a series of SSD and MIP images. The connectivity and opacity transfer function for the VR were determined so as to maximize the vascular visualization while minimizing the non-vascular display. Visualization of VIE images was based on CT number thresholding, and the details of generating the VIE images have been described elsewhere (9). The CVR was generated by setting the line at the centre of the abdominal aorta in order to produce coronal or sagittal images for demonstrating the stent graft and the aortic branches. Technical details for the generation of these 3D reconstructions have been described elsewhere (12).
3D Visualization and the Diagnostic Performance of Multislice CT
3D reconstructions were successfully generated for all the patients. The follow-up time ranged from 24 months to 54 months, with a mean period of 40 + 7.6 months. Endoleaks were found in five cases, with type I and III in one patient each, respectively, and type II in three cases. The renal function was evaluated by measuring the serum creatinine levels and it was not significantly affected in all cases, except in one patient who developed chronic renal failure due to an atrophic left renal artery and that patient received renal dialysis. Distal stent graft migration occurred in four patients and their treatment was under observation at the current follow-up.
Surface Shaded Display (SSD)
The SSD represents the surface of a structure within a volume dataset. It is quickly generated because it relies on simple thresholding. It has superiority for the speed and flexibility of rendering images for the visualization of the stent graft and the arterial branches (Fig. 1A). However, the surface of the is derived from only a small proportion of the available data (less than 10%), and it is prone to artifacts (Fig. 2A), and this is especially apparent in the patients treated with stent graft due to the high density of the metal wires. Thus, calcification in the aortic wall and the configuration of the aneurysm sac cannot be accurately assessed on the SSD images when compared to the corresponding MIP images (Figs. 1B, 2B). SSD has only a small role to play in the follow-up of aortic stent grafting as no additional information can be obtained from it when compared to the axial 2D images (8).
Maximum Intensity Projection (MIP)
The MIP was found to be valuable for post stent grafting as it creates angiographic-like images that are similar to those from conventional angiographic examinations. High density calcification, contrast-enhanced vessels, aortic stent wires, the 3D relationship of the aortic stent graft and the arterial branches can be clearly visualized on the MIP images (Figs. 1B, 2B). The MIP images were also found to be useful for assessment of stent graft migration at the regular follow-ups, and they were reported to be more accurate than the axial 2D images (Fig. 3) (13). One of the disadvantages of MIP generation is having to remove the bony component from the volume data, which is a time-consuming procedure. However, development of computer software allows the user to automatically remove bony structures and the processing time has been significantly decreased, which makes MIP image processing an acceptable technique in routine clinical practice.
Curvilinear Reformation (CVR)
Most of the abdominal aortic aneurysms are angulated to a variable extent and they have a curved path along the abdominal aorta, so conventional multiplanar reformation is not always able to reveal the anatomical information required for assessment. Curvilinear reformation allows for generating images with the desired anatomy by segmenting the anatomical structure with a curved central lumen line (Fig. 4). However, caution should be taken into account when choosing the selection plane as the line must be set in the centre of anatomical structures, otherwise, not all of the information will be displayed in the reformatted images and some of the information will be missed (Fig. 5). Therefore, a series of CVR images instead of individual ones are required for accurately evaluating aortic stent grafting, which is the main limitation of the CVR.
Volume Rendering (VR)
VR was able to demonstrate all the anatomical structures such as the stent graft and aortic branches because it uses all of the information contained inside a volume dataset. Moreover, each anatomical structure can be coded with a specific colour and opacity value for every attenuation value of the CT data (Fig. 6A). This allows observers to easily identify a particular structure within the volume dataset and it clearly demonstrates the 3D relationship between the stent graft and the arterial branches (Figs. 6B, C ). Because the generation of VR images relies on the threshold selection and there is no requirement to remove unwanted structures, the processing time is very short, i.e., only a few minutes (14). The main limitation of VR is that the final image quality depends on the segmentation of the volume dataset that is used by the rendering algorithm (Fig. 7). Therefore, acquisition of the optimal source of CT data is of utmost importance to ensure the VR image quality.
Virtual Intravascular Endoscopy (VIE)
Unlike other 3D reconstructions, VIE provides unique intraluminal information about the stent wires relative to the arterial branch ostia. The number of stent wires crossing the arterial ostia and the configuration of encroachment can be clearly visualized on VIE (Fig. 8). The VIE images can enhance our understanding of the effect of the aortic stent graft on the arterial branch ostia, especially in the patients with AAA that is treated with a suprarenal stent graft (8, 9). Our experience has shown that VIE could be a valuable visualization technique for the follow-up of aortic stent grafting. Comparison of the VIE images at regular follow-up also allows clinicians to assess the effect of stent wires on the arterial ostia, or we can observe the stent wires regarding the distal migration or the reduction of the cross-sectional area of the aortic ostia (Fig. 9) (13). One of the limitations of VIE visualization is that it is a time-consuming procedure, which takes about 30 minutes. However, with acquiring more experience, the time required for generating VIE can been greatly reduced. Another limitation is that the stent wire thickness is overestimated on VIE images due to the point spread function, which makes a stent wire looks thicker that it really is (Fig. 9). Thus, the actual coverage of the stent wires to the aortic ostia is hard to estimate, although we know that only a small proportion of the aortic ostium is covered (15).
DISCUSSION and CONCLUSION
With the introduction of MSCT scanners, CT angiography of the aortic stent graft is becoming a more important procedure. In contrast to single slice CT, CTA can be performed more efficiently with MSCT scanners because of the faster scanning speed and the higher spatial and temporal resolution. In an early study, Rubin et al. (7) showed that performing CTA with a four-detector row CT scanner was faster, and the scanning was possible with thinner collimation and a reduced dose of contrast medium. However, the studies using 16 slice CT scanners with a submillimeter section thickness demonstrated higher quality CT angiograms than did those obtained from 4 slice scanners for the evaluation of aortoiliac and lower extremity arteries (16, 17). With the recent introduction of the 64 slice scanner, spiral scanning with even a shorter time represents a further leap in improving the spatial and temporal resolution for routine clinical applications of CTA (18, 19). Therefore, for the patients undergoing stent graft treatment, CT angiographic examinations with submillimeter resolution in the pure arterial phase will become feasible even for an extended anatomic range. Two of our patients' volume source data were suboptimal and the image quality of VR was affected, as is shown in Figure 7. For one patient, the CT attenuation was measured to be lower than 200 HU at the level of the renal artery. Although adequate contrast enhancement was obtained in another patient (CT attenuation of 230 HU), the image quality was still affected due to the low signal to noise ratio (increased image noise). We believe that the improved longitudinal resolution with a 64 slice scanner will overcome these limitations that we encountered in our study.
3D CT postprocessing and reconstruction has become a part of the clinical protocol and this has been widely used as an effective alternative to conventional angiography for preoperative planning and the post-operative follow-up of aortic stent grafts (5, 20). 3D CT angiography is our routine imaging modality to obtain the essential measurements for planning aortic stent graft placement, to assess the patency of stent-covered renal arteries and to detect endoleaks for post stent grafting follow-up (8). Our results have demonstrated the potential value of 3D MSCT reconstructions, including MIP, VR and VIE, which provide clinicians with additional information for evaluating the effects of endovascular repair. MIP clearly shows the aortic stents and the arterial branches, and it was more accurate for measuring the stent graft migration than were the conventional 2D images (13), while VR is an efficient visualization tool for demonstrating the 3D relationship between the stent graft and the arterial branches. As the long-term effects of aortic stent grafting are unclear, and especially the effects of transrenal fixation, the additional information provided by VIE is considered valuable for surgeons to assess the outcomes of transrenal fixation in the following two settings: potential interference with renal blood flow by demonstrating the encroachment of stent wires to the renal artery ostia, which may alter the haemodynamics and lead to the dispersion of late multiple emboli (21), or any reduction of cross-sectional area of the aortic branch ostia that's caused by the presence of stent wires (15), which may affect the physiologic change of renal function such as renal perfusion. The diagnostic value of CT VIE in aortic stent grafting has been discussed in detail elsewhere (8, 9, 13, 15).
In conclusion, based on our experience, 3D CT reconstructions offer additional information when compared to the 2D axial images in post-stent grafting. Reliable recognition of the diagnostic performance of each reconstruction is of paramount importance for clinicians to effectively utilize the MSCT imaging technique as some of the 3D postprocessings are time-consuming such as occurs with MIP and VIE. VR is the most efficient reconstruction method for visualizing the aortic stent graft relative to the arterial branches, whereas VIE is a valuable tool for visualizing intraluminal encroachment of the stent wires to the aortic ostia.
XML Download