Clinical & Experimental Cardiology
Open Access

Influence of C-V Model on Coronary Angiogram and the Vasculature

Hajja Ibrahim^{* *}Correspondence: Hajja Ibrahim, Department of Cardiac Surgery, International University of Africa, Khartoum, Sudan, Email:

Author info »

Description

The "gold standard" for diagnosing coronary heart disease is coronary angiography, for which vessel segmentation and identification technologies are of great importance. However, there are several challenges in these investigations due to the peculiarities of coronary angiograms, such as the complicated and varied morphology of coronary artery anatomy and the noise induced by many variables. In order to solve these issues, we develop a preprocessing method that combines blockmatching and 3D filtering, unsharp masking, contrast-limited adaptive histogram equalization, and mu ltiscale picture enhancement. This method enhances the vascular structure while also improving the image quality. We extract the vascular contour from the C-V model in order to perform vessel segmentation. Finally, in order to achieve automatic identification of the vascular skeleton, we provide an enhanced adaptive tracking approach. According to our tests, the preprocessing scheme effectively highlights vascular structures and limits background, the C-V model accurately extracts the continuous vessel contour, and it is confirmed that the proposed tracking method is more accurate and robust than the current adaptive tracking method [1]. Currently, cardiovascular disease is acknowledged as one of the major chronic illnesses contributing to fatalities worldwide. Cardiovascular diseases have been leading all other diseases in terms of morbidity and mortality during the past several years. A frequent and reliable technique for detecting coronary heart disease is Coronary Angiography (CA). It is frequently used in clinical diagnosis and is known as the "gold standard" for the diagnosis of coronary heart disease.

Human arteries and vessels are often X-ray invisible. However, the arteries and vessels can be seen by injecting X-ray impermeable chemicals into the coronary arteries and then exposing the coronary artery area to X-rays [2,3]. Doctors must independently determine the location and severity of coronary artery stenosis based on the picture in order to choose the best course of treatment. However, this approach inevitably results in a lot of repeated labour and subjective inaccuracies. Therefore, developing systems that can segment and identify arteries in angiograms is quite beneficial. Numerous academics have suggested numerous approaches because of this.

Image segmentation has long been a key area of interest in image processing. Numerous segmentation technologies for vessels have been put out up until this point. We can categorize vessel segmentation technologies into three groups based on the two properties of discontinuity between regions and similarity within regions: boundary-based segmentation technologies, regionbased segmentation technologies, and technologies combined with specific theories and tool segmentation.

By examining the characteristics of the eigenvalues of the Hessian matrix of spherical, tubular, and sheet-like structures at a specific scale, a multi-parameter similarity function for improving vessels was devised. A novel active contour C-V model to evolve the curve through the minimization of the energy function based on the simplified Mumford-Shah model and the level set concept. Deep learning techniques have lately been widely applied to the segmentation of vessels. As an illustration, great segmentation accuracy was attained when the 3D U-Net was trained to conduct three-dimensional vascular segmentation.

Various techniques for vascular identification have also been researched, including multiscale-based and tracking-based techniques. The tracking-based approach has proven to be the most successful of these. Without scanning the full image, it can identify coronary information based on the angiogram's local reaction. Because the seed locations were manually set during the coronary artery extraction process, the extraction outcome is unstable [4]. To address this issue, a ridge point detection-based automatic seed point acquisition approach was developed. These ridge points act as seed locations for the adaptive tracking of the coronary arteries centre-line. A ridge detection-based multi-scale spatial centre-line tracking approach that extracts the ridge from the Hessian matrix using its eigenvalue decomposition. However, there is still potential for development in terms of these systems' accuracy and robustness because of restrictions in algorithm design and the consequences of poor image quality, noise, etc.

Our primary efforts and contributions were as follows: first, we created a pre-processing method to improve the image's quality and highlight the vascular structure [5]. After that, we segmented the vessels using the C-V model. To achieve automatic identification of the vascular skeleton, we lastly proposed an enhanced adaptive tracking technique. According to our trials, this approach produced better results than previous ones.

Conclusion

The ability to segment and automatically identify arteries in coronary angiograms is made possible by an image preprocessing method that made use of the C-V model and offered an enhanced adaptive tracking technique. One of these techniques, the improved adaptive tracking method, comprises our key improvements that can improve the ability to spot vessels. The proposed method has been the subject of numerous tests, and the findings show that our method is more reliable and precise than the previous way. Traditional image processing techniques fall short due to the coronary angiograms' intricacy, as was previously discussed. The tracking algorithm will therefore continue to be improved in the upcoming work, and deep learning image processing research will be conducted.

References

1. Zamani Boroujeni F, Rahmat RW, Mustapha N, Affendey LS, Maskon O. Coronary artery center-line extraction using second order local features. Comp Math Method Med. 2012.

   [Cross Ref][Google Scholar]

2. Jain K, Kumar A. An optimal RSSI-based cluster-head selection for sensor networks. Int J Adapt Innov Syst. 2019;2(4):349-361.

   [Google Scholar]

3. Shoujun Z, Jian Y, Yongtian W, Wufan C. Automatic segmentation of coronary angiograms based on fuzzy inferring and probabilistic tracking. Biomed Eng Online. 2010;9(1):1-21.

   [Cross Ref][Google Scholar]

4. Xiao R, Yang J, Ai D, Fan J, Liu Y, Wang G, et al. Adaptive ridge point refinement for seeds detection in x-ray coronary angiogram. Comput Math Methods Med. 2015;2015:502573.

   [Cross Ref][Google Scholar][Pub Med]

5. Dabov K, Foi A, Katkovnik V, Egiazarian K. Image denoising by sparse 3-D transform-domain collaborative filtering. IEEE Trans Image Process. 2007;16(8):2080-2095.

   [Cross Ref][Google Scholar]