Segmentation of Carotid Arteries in CTA Images using Region-based Active Contours and Classification


BOZKURT F., KÖSE C., SARI A.

2017 International Artificial Intelligence and Data Processing Symposium (IDAP), Malatya, Türkiye, 16 - 17 Eylül 2017 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/idap.2017.8090261
  • Basıldığı Şehir: Malatya
  • Basıldığı Ülke: Türkiye
  • Anahtar Kelimeler: Carotid Artery Segmentation, Vessel segmentation, Region Based Active Contour Segmentation, Classification, LUMEN SEGMENTATION, ANGIOGRAPHY
  • Atatürk Üniversitesi Adresli: Evet

Özet

Carotid artery stenosis is usually the bottleneck caused by atherosclerosis or carotid artery luminal narrowing. Carotid arteries are in close proximity to bone and bony structures as spongy. Contrast-enhanced computerized tomography angiography (CTA) is used to monitor and measure carotid arteries under the control of an expert. Recently, there is a strong and growing demand for improving the computer aided carotid segmentation process. Recently, there has been a strong and growing demand for computer aided carotid segmentation. In this study, segmentation of the vessels in the CTA images is performed by using region-based active contours method and classification of the segmented regions. The boundaries of the vessel-bone regions are found by region-based active contour segmentation. After segmentation of regions with high gray-scale level values, such as veins and bones, these regions should be separated from each other. In order to perform only carotid artery segmentation in CTA images, it is necessary to eliminate bone fragments and noisy vessel-like structures. For this purpose, in this study, a decision-making mechanism at the point of vein-bone separation is established to classify the segmented regions with a supervised learning system. The method is applied on different patients' CTA images and the performance evaluation is done with statistical and area-based metrics. In these experimental results; average of over 89% Dice similarity 99% accuracy are obtained.