Coronary Vessel Segmentation by Coarse-to-Fine Strategy using Otsu Algorithm and Decimation-Free Directional Filter Bank

Abstract

In this study, a novel hierarchical approach is investigated to extract coronary vessel from X-ray angiogram. First, we propose to combine Decimation-free Directional Filter Bank (DDFB) and Homographic Filtering (HF) in order to enhance X-ray coronary angiographic image for segmentation purposes. Because the blood vessel ensures that blood flows in only one direction on vessel branch, the DDFB filter is suitable to be used to enhance the vessels at different orientations and radius. In the combination with HF filter, our method can simultaneously normalize the brightness across the image and increases contrast. Next, a coarse-to-fine strategy for iterative segmentation based on Otsu algorithm is applied to extract the main coronary vessels in different sizes. Furthermore, we also propose a new approach to segment very small vessels. Specifically, based on information of the main extracted vessels, we introduce a new method to extract junctions on the vascular tree and level of nodes on the tree. Then, the window based segmentation is applied to locate and extract the small vessels. Experimental results on our coronary X-ray angiography dataset demonstrate that the proposed approach can outperform standard method and attain the accuracy of 71.34%.

Fig. 1. An illustration of the proposed method for the coronary vessel segmentation.

Fig. 2. Our pre-processing process.

Fig. 3. An example of diffusion on our atabase. a) Input image. b) Boundary between background and foreground based edge detector. c) Circle mask based on Hough Transform d) Diffused result from input.

Fig. 4. Large vessel extraction based coarse-to-fine segmentation strategy.

Fig. 5. An block diagram of the vessel enhancement based on the DDGFB and HF filter.

Fig. 6. An example of coarse-to-fine segmentation strategy based on Otsu algorithm to extract the large vessels. a) Input image, b) Segmentation result.

Fig. 7. A block diagram of our method for small vessel extraction.

Fig. 8. An example of connecting the nearest the center-lines. a) Input image, b) Image after connecting the nearest the center-lines

Fig. 9. An example of tracing the vessel branch and junction. a) The seed point, b) Tracking candidate points and the junction

Fig. 10. An example of eliminating the invalid branches. In this case, the branches corresponding to directions $\overrightarrow{v_1}$ and $\overrightarrow{v_3}$ are eliminated.

Fig. 11. An example of the level of the vessel branch and degree of nodes.

Fig. 12. An example of the level of the vessel tree and degree of nodes on our dataset

Algorithm 2. Small vessel detection and segmentation based on window analysis.

Fig. 15. Small vessel extraction. a) Part of input image, b) Result from large vessel extraction approach, c) Result from small vessel extraction approach

Fig. 16. An image IM-0001-000130 in our dataset.

Fig. 17. Ground truth of the IM-0001-000130 image.

Fig. 18. Segmentation result of the proposed approach on our dataset. a) Input image, b) Segmented vessels, c) Ground truth

Fig. 19. Some false alarms of the proposed approach on our dataset. a) Input image b) False alarms from segmented result, c) Ground truth

Fig. 20. Comparison of the segmentation results on our dataset a) Input image, b) Our result, c) Result in [18]

Fig. 13. An example of window constructed between two nodes.

Fig. 14. An example of contrast enhancement in the window analysis. a) Input image, b) Result from contrast enhancement

Table 1. The optimal parameters for the DDFB in our experiments.

Table 2. Comparison performance of the coronary vessel segmentation on our dataset.

Algorithm 1. Branch and junction detection