Tools
Tools
Ma, YingLiang 
ORCID: https://orcid.org/0000-0001-5770-5843, Alhrishy, Mazen, Panayiotou, Maria, Narayan, Srinivas Ananth, Fazili, Ansab, Mountney, Peter and Rhode, Kawal S.
(2017)
Real-time guiding catheter and guidewire detection for congenital cardiovascular interventions.
In:
Functional Imaging and Modelling of the Heart.
Lecture Notes in Computer Science
.
Springer, 172–182.
ISBN 9783319594477
Abstract
Guiding catheters and guidewires are used extensively in pediatric cardiac catheterization procedures for congenital heart diseases (CHD). Detecting their positions in fluoroscopic X-ray images is important for several clinical applications, such as visibility enhancement for low dose X-ray images, and co-registration between 2D and 3D imaging modalities. As guiding catheters are made from thin plastic tubes, they can be deformed by cardiac and breathing motions. Therefore, detection is the essential step before automatic tracking of guiding catheters in live X-ray fluoroscopic images. However, there are several wire-like artifacts existing in X-ray images, which makes developing a real-time robust detection method very challenging. To solve those challenges in real-time, a localized machine learning algorithm is built to distinguish between guiding catheters and artifacts. As the machine learning algorithm is only applied to potential wire-like objects, which are obtained from vessel enhancement filters, the detection method is fast enough to be used in real-time applications. The other challenge is the low contrast between guiding catheters and background, as the majority of X-ray images are low dose. Therefore, the guiding catheter might be detected as a discontinuous curve object, such as a few disconnected line blocks from the vessel enhancement filter. A minimum energy method is developed to trace the whole wire object. Finally, the proposed methods are tested on 1102 images which are from 8 image sequences acquired from 3 clinical cases. Results show an accuracy of 0.87 ± 0.53 mm which is measured as the error distances between the detected object and the manually annotated object. The success rate of detection is 83.4%.
| Item Type: | Book Section |
|---|---|
| Faculty \ School: | Faculty of Science > School of Computing Sciences |
| UEA Research Groups: | Faculty of Science > Research Groups > Norwich Epidemiology Centre Faculty of Medicine and Health Sciences > Research Groups > Norwich Epidemiology Centre Faculty of Science > Research Groups > Data Science and AI |
| Related URLs: | |
| Depositing User: | LivePure Connector |
| Date Deposited: | 06 Jan 2023 11:31 |
| Last Modified: | 10 Dec 2024 01:12 |
| URI: | https://ueaeprints.uea.ac.uk/id/eprint/90442 |
| DOI: | 10.1007/978-3-319-59448-4_17 |
Actions (login required)
 |
View Item |