Abstract

Background Early detection of small type I endoleaks after endovascular aneurysm sealing is mandatory because they can rapidly progress and lead to severe complications. Recognition of endoleaks can be challenging due to the appearances on computed tomography unique to endovascular aneurysm sealing. We aimed to validate the accuracy and added value of subtraction computed tomography imaging using a post-processing software algorithm to improve detection of endovascular aneurysm sealing-associated endoleaks on postoperative surveillance imaging. Methods The computed tomography scans of 17 patients (16 males; median age: 78, range: 72-84) who underwent a post-endovascular aneurysm sealing computed tomography including both non-contrast and arterial phase series were used to validate the post processing software algorithm. Subtraction images are produced after segmentation and alignment. Initial alignment of the stent segmentations is automatically performed by registering the geometric centers of the 3D coordinates of both computed tomography series. Accurate alignment is then performed by translation with an iterative closest point algorithm. Accuracy of alignment was determined by calculating the root mean square error between matched 3D coordinates of stent segmentations. Results The median root mean square error after initial center of gravity alignment was 0.62 mm (IQR: 0.55-0.80 mm), which improved to 0.53 mm (IQR: 0.47-0.69 mm) after the ICP alignment. Visual inspection showed good alignment and no manual adjustment was necessary. Conclusions The possible merit of subtraction computed tomography imaging for the detection of small endoleaks during surveillance after endovascular aneurysm sealing was illustrated. Alignment of different computed tomography phases using a software algorithm was very accurate. Further studies are needed to establish the exact role of this technique during surveillance after endovascular aneurysm sealing compared to less invasive techniques like contrast-enhanced ultrasound.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call