Output details

Automatic Segmentation of Coronary Arteries in CT Imaging in the Presence of Kissing Vessel Artifacts

Anatomically-correct vessel segmentation, providing quantitative vessel analysis, is essential in diagnosing vascular disease. A new active contour-based algorithm, utilizing global and local intensity statistics for coronaries segmentation in CTA is proposed. This approach accommodates for uneven brightness distributions over the image, while escaping local minima. Moreover, the problem of kissing vessel artefacts, a partial volume effect, which may result in artificial vessel junctions, thus distorting the vessel’s true geometry is addressed. A novel slice-by-slice correction scheme, based on multi-region competition is used, thus alleviating the need for clinicians to manually remove such artefacts, necessary in cardiovascular assessment.