Iterative algorithm for metal artefact reduction (iMAR) mimicking total aortic stent graft thrombosis in a patient after endovascular abdominal aneurysm repair (EVAR)

A patient with a medical history of atherosclerotic ulcer in an infrarenal abdominal aortic aneurysm treated by endovascular aneurysm repair (EVAR), and with a recent history of acute-on-chronic pancreatitis, presented with renewed acute severe upper abdominal pain.

For further evaluation, the patient underwent contrast-enhanced abdominal computed tomography (CT).

CT identified a pre-existing cystic formation in the pancreatic tail, most probably corresponding to a pancreatic pseudocyst (not shown). Otherwise, the pancreatic parenchyma appeared morphologically unaltered, with no swelling or necrosis. Furthermore, there was no evidence of extra-pancreatic fluid collections or abscess in the upper abdomen.

After initial reconstruction with an iterative algorithm for metal artefact reduction (iMAR), the lumen of the EVAR stent graft appeared homogeneously hypodense, indicating a lack of endo-luminal contrast. However, after a second reconstruction of the source data set without iMAR, the stent lumen appeared homogeneously hyperdense, confirming normal perfusion through the stent-graft. In the reconstructed scan without iMAR, more intense metal artefacts overlaid the tissue surrounding the stent-graft.

Background

While most cases of acute pancreatitis are mild, challenges remain when managing severe cases [1], especially because complications can be lethal, such as with pancreatitis-induced aortic thrombosis [2]. Patients after EVAR have an additional risk of thrombotic occlusion [3]. The metal mesh of some grafts may cause imaging artefacts, which can limit the diagnostic value, especially when assessing the stent lumen. iMAR is a vendor-specific dedicated reconstruction algorithm that may help reduce these limitations [4]. It is based on a sinogram inpainting approach with linear interpolation to reduce metal artefacts in the standard weighted filtered back-projection [5]. A metal-only image is calculated by threshold-segmentation and then again forward projected to identify and replace corrupted projections with interpolated data from adjacent projections [6]. However, iterative reconstruction algorithms themselves may limit the diagnostic value of the reconstructed images because of reconstruction artefacts or image distortion [7]. Two major artefacts have been reported in iMAR: new streak artefacts that occur between interpolated and non-interpolated data and blurring caused by data that cannot be recovered by interpolation [8].

Clinical perspective

Upper abdominal pain is a typical clinical presentation of acute pancreatitis. However, this sign is highly nonspecific and cannot easily be distinguished from other causes of abdominal pain, especially acutely. A potential differential diagnosis is visceral ischaemia; for example, due to total aortic branch occlusion. The correct diagnosis of these entities is crucial as immediate intervention may be required. In case of an image artefact, however, no intervention would be necessary.

Imaging perspective

Given the lack of endo-luminal contrast during the initial reconstruction with iMAR, total aortic stent graft thrombosis was a severe differential diagnosis. The smaller vessels in the immediate vicinity ventral to the stent-graft were properly contrasted, which argued against an artefact. However, the lack of contrast was limited to the levels of coated mesh and strictly followed the image orientation, showing no concavity or convexity, which is rather atypical for a thrombus. Moreover, the aortic lumen was properly contrasted distal to the stent-graft without clearly visible collaterals. Finally, image reconstruction of the source data set without iMAR revealed regular endo-luminal contrast, confirming that the initial endo-luminal hypodensity with iMAR was an artefact.

Outcome

To the best of our knowledge, this was the first reported case in which metal artefact reconstruction algorithms may have led to a total loss of endo-luminal contrast in a stent-graft. This artefact could be specific to certain stent-graft models and should be further investigated to avoid unnecessary interventions.

Take home message

Reconstruction artefacts should be considered as a possible cause of apparent total stent occlusion in iteratively-reconstructed CT images when the medical history or clinical picture is inconclusive. In particular, radiologists with less experience should be aware of the existence of these artefacts.