Low-dose 2D X-ray angiography enhancement using 2-axis PCA for the preservation of blood-vessel region and noise minimization. Issue 123 (January 2016)

• Record Type: Journal Article
• Authors: Lee, Yong Geun; Lee, Jeongjin; Shin, Yeong-Gil; Kang, Ho Chul
• Is Part Of: Computer methods and programs in biomedicine. Issue 123(2016)
• Page Start: 15
• Page End: 26
• Publication Date: 2016-01
• Subjects: Vessel enhancement -- 2-Axis PCA -- 2D angiography -- Noise reduction -- Vessel segmentation -- Energy compaction; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2015.09.011
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: A novel and fast vessel-enhancement method for 2D angiography method is proposed. The results showed that we obtain nature enhanced X-ray images using our method and the processing time is very small. Our method is applicable to 2D angiography procedures such as catheterization, which requires rapid and natural vessel enhancement. Abstract: Enhancing 2D angiography while maintaining a low radiation dose has become an important research topic. However, it is difficult to enhance images while preserving vessel-structure details because X-ray noise and contrast blood vessels in 2D angiography have similar intensity distributions, which can lead to ambiguous images of vessel structures. In this paper, we propose a novel and fast vessel-enhancement method for 2D angiography. We apply filtering in the principal component analysis domain for vessel regions and background regions separately, using assumptions based on energy compaction. First, we identify an approximate vessel region using a Hessian-based method. Vessel and non-vessel regions are then represented sparsely by calculating their optimal bases separately. This is achieved by identifying periodic motion in the vessel region caused by the flow of the contrast medium through the blood vessels when viewed on the time axis. Finally, we obtain noise-free images by removing noise in the new coordinate domain for the optimal bases. Our method was validated for an X-ray system, using 10 low-dose sets for training and 20 low-dose sets for testing. The results were compared with those for a high-dose dataset with respect to noise-free images. The average enhancement rate was 93.11 ± 0.71%. The average processing time for enhancing video comprising 50–70 frames was 0.80 ± 0.35 s, which is much faster than the previously proposed technique. Our method is applicable to 2D angiography procedures such as catheterization, which requires rapid and natural vessel enhancement.