Presenting a method for right and left ventricle segmentation from cardiac MRI images

Number of pages: 103 File Format: word File Code: 32139
Year: Not Specified University Degree: Master's degree Category: Electronic Engineering
  • Part of the Content
  • Contents & Resources
  • Summary of Presenting a method for right and left ventricle segmentation from cardiac MRI images

    Master thesis in the field

    Medical-Bioelectric Engineering

    Abstract

    Providing a method for right and left ventricle segmentation from cardiac MRI images

     

    Examining the structure and function of heart ventricles in magnetic resonance images is considered an important step in the management of many heart disorders. Although the manual segmentation of heart ventricles achieves good results, manual segmentation of ventricles, especially the right ventricle, is a time-consuming task due to its complex geometry. Therefore, automatic segmentation of the right and left ventricles seems necessary. In this project, a new automatic method is presented based on the combination of particle swarm optimization (PSO) and improved random search. PSO is an evolutionary method based on the population, through this method, we segment parts of the image with similar brightness, and finally, the final segmentation of the right and left ventricles is done by the improved random scanner. In the random survey method, a number of pixels are labeled by the user, but in the presented method, the selection of labeled points is done automatically and by the PSO method.

    To check the accuracy of the presented method, we applied the proposed method on a large number of different images and we observed acceptable results from a clinical and technical point of view.

    key words: magnetic resonance images, segmentation of left and right ventricles, particle cluster optimization method, random navigation algorithm.

    1-1- Introduction

    The heart is a vital part of the circulatory system of the human body. Proper functioning of the heart is necessary to prevent cardiovascular diseases. Lack of exercise, inactivity, stress, improper diet, and genetic factors all play a fundamental role in causing and increasing cardiovascular disorders. Cardiovascular diseases are currently one of the top three causes of human death and disability worldwide and are becoming the main cause of death and disability in most countries [1]. Therefore, the control and treatment of these diseases is an important issue. In the past, blood pressure monitoring, blood tests to detect cholesterol and ECG were the methods available to the doctor to monitor the cardiovascular health of people. In recent years, with advances in medical science and medical engineering, imaging equipment has come to help doctors to diagnose and control diseases. As far as the diagnosis and treatment of cardiovascular diseases rely to a large extent on various imaging methods such as echocardiography, computed tomography (CT), coronary angiography and magnetic resonance imaging (cardiac MRI).

    The high definition of the image is the main advantage of MRI, which has the ability to image the soft tissues of the body. Therefore, MRI is very suitable for imaging the heart, brain, muscles and tumors. Multimodal imaging is another One of the characteristics of MRI is that it is able to take cross-sectional images on any surface, without changing the patient's position. Also, because MRI uses electromagnetic radiation (RF), it does not have the harmful effects of radioactivity and X-ray radiation in nuclear and CT imaging. Performing MRI on a large number of cardiovascular patients in the community produces a significant amount of data The doctor's work is time-consuming. Also, their interpretation relies on the doctor's judgment and therefore can be prone to errors. Therefore, the use of computer and image processing methods has been proposed as a solution to these problems. In the last few decades, the analysis of cardiac images, especially the segmentation [1] of cardiac images, has been the subject of many studies. Segmentation of images is considered a prerequisite for many high-level tasks such as image analysis, computer-aided diagnosis, geometric modeling of anatomical structures, or the construction of biomechanical models that are used to simulate surgery.

    Segmentation of heart images is to identify the heart or any of its anatomical or physiological characteristics from two-dimensional or three-dimensional images.

    The main goal in image segmentation is to segment the given image into two or more regions. A simple example of heart image segmentation in two areas is shown in Figure 1-1, area u1 corresponds to the desired object (foreground) and area u2 is the background of the image.

    Figure 1-1. Image segmented into two areas

    In other words, the goal in this example is to identify two groups of pixels u1 and u2, which are defined as follows.

    = u1

    (1-1)

    = u2

    Many studies have been done on automatic heart segmentation methods, but the environment and contour detection still need to be improved to the extent of manual methods. Despite the many works that have been done on segmenting the right and left ventricles of cardiac images, the problem still remains [2].

    There are challenges in segmenting the right and left ventricles of cardiac images, which are mainly due to the anatomy of the heart and the characteristics of MRI images. Some of the problems that exist in the segmentation of heart images are:

    Variation in speed, position, rotation of the heart and differences in contrast [2] and resolution of the images, which can be problematic for any of the segmentation methods.

    Reconstructed images may include noise and other artifacts that are mostly caused by the following reasons:

    Noise Thermal

    Heart movement and breathing

    A small error, which occurs due to the estimation of the continuous borders of the heart by a digital curve.

    Tissues that exist in the heart and adjacent to the heart may have similar gray level brightness.

    The right ventricle, unlike the left ventricle, which has a regular shape And it has an oval, crescent-shaped irregular shape, whose wall thickness is 3 to 6 times thinner than the left ventricular cavity (Figure 1-2), which leads to the limitation of accuracy in the spatial resolution of MRI images. Geometric shape of left and right ventricle

    Due to the mentioned reasons, the problem of dividing the ventricles, especially for the right ventricle, still remains. Due to the regularity of the shape of the left ventricle and because the function of the left ventricle is more vital than the right ventricle, most of the studies have focused on the left ventricle, but in recent years, methods have also been presented to segment the right ventricle. style="direction: rtl;"> 

    Segmentation of the Right and Left Ventricles in Cardiac Mri

     

    By

    Maliheh Sehati

     

    Evaluation of cardiac ventricular structure and function in magnetic resonance (MR) images is an important step in the management of most cardiac disorders. Although the manual segmentation of the cardiac ventricle produces good results, but manual segmentation of the cardiac ventricles is time-consuming, especially for the right ventricle (RV) due to its complex geometry, therefore automatic segmentation of the RV and LV is important. In this study, we propose a novel approach for automatic segmentation of the right and left ventricle in cardiac MR images based on the combination of Particle Swarm Optimization (PSO) and improved random walks algorithm.

  • Contents & References of Presenting a method for right and left ventricle segmentation from cardiac MRI images

    List:

    Chapter One: Introduction

    1-1- General.. 2

    1-2- Human heart.. 7

    1-2-1- Structure and function of the heart. 7

    1-3-MRI imaging. 10

    1-3-1- Cardiac MRI. 12

    1-4- Justification of the necessity of carrying out the plan and work method. 14

    1-5-Research issue from a medical point of view. 16

     

    Chapter Two: Subject and Research Background

    2-1-Introduction.. 18

    2-2- Cardiac MRI image segmentation methods. 18

    2-2-1- Automatic segmentation method. 20

    2-2-2- semi-automatic methods. 22

    2-2-2-1- Division with poor or no knowledge. 22

    2-2-2-1-1- Image-based methods. 22

    2-2-2-1-2- methods based on pixel classification. 23

    2-2-2-1-3- variable models. 24

    Title

    2-2-2-1-4 Conclusion. 26

    2-2-2-3- Division with strong knowledge. 27

    2-2-2-3-1- Transforming the model with strong prior knowledge. 28

    2-2-2-3-2- active form and appearance models. 28

    2-2-2-3-3- segmentation based on atlas. 30

    2-2-2-3-4- conclusion. 32

     

    Chapter three: Right and left ventricle segmentation from heart MRI images

    3-1- Introduction.. 38

    3-2- PSO method.. 40

    3-3- Structural operation. 44

    3-4- Random browser method. 47

    3-4-1- The weight of the edges. 50

    3-4-2- Combined Dirichlet problem. 51

    3-4-3- orbital analogy. 51

    3-4-4- The connection between the method and the diffusion process in machine vision. 52

    3-4-5- improved random navigation method. 54

    3-4-6- Summary of the algorithm. 55

    3-4-7- Features of the algorithm in theory. 55

    3-4-8- behavioral characteristics. 57

    3-4-8-1- Weak borders. 57

    3-4-8-2- noise resistance. 58

    3-4-8-3- Ambiguous and unlabeled areas. 59

     

    Title                                                                                                                                                                       page

    Chapter Four: Examining the results

    4-1- Introduction.. 61

    4-2- Characteristics of the data. 61

    4-3- How to implement the proposed method. 62

    4-4- Discussion on the results of the proposed methods. 64

    4-5- Technical review. 67

    4-5-1- Dice coefficient. 69

    4-5-2- Calculation of similarity. 70

    4-6- Comparison with previous methods. 71

    4-7- Conclusion.. 76

    Chapter Five: Conclusion and Future Works

    5-1-Introduction.. 78

    5-2- Suggestions for future studies. 79

     

    List of references

    Source:

     

    [1] Lloyd-Jones, D., Adams, R. J., Brown, T. M., Carnethon, M., Dai, S., De Simone, G., and Wylie-Rosett, J. (2010). Heart disease and stroke statistics, A report from the American Heart Association. Circulation, 121(7), e46-e62.

    [2] Petitjean, C., Dacher, J. N. (2011). A review of segmentation methods in short axis cardiac MR images. Medical image analysis, 15(2), 169-184.

    [3] Shors, S. M., Fung, C. W., François, C. J., Finn, J. P., and Fieno, D. S. (2004). Accurate Quantification of Right Ventricular Mass at MR Imaging by Using Cine True Fast Imaging with Steady-State Precession: Study in Dogs 1. Radiology, 230(2), 383-388.

    [4] Ghose, S. (2009). Deformable Model based Computation of Ejection Fraction of Right Ventricle from Cine MRI (Doctoral dissertation, A Thesis Submitted for the Degree of M. Sc. Erasmus Mundus in Vision and Robotics, (VIBOT), University of Bourgogne). [5] Alfakih, K., Plein, S., Thiele, H., Jones, T., Ridgway, J. P., and Sivananthan, M. U. (2003). Normal human left and right ventricular dimensions for MRI as assessed by turbo gradient echo and steady-state free precession imaging sequences. Journal of Magnetic Resonance Imaging, 17(3), 323-329.

    [6] Rougon, N., Petitjean, C., Prêteux, F., Cluzel, P., and Grenier, P. (2005). A non-rigid registration approach for quantifying myocardial contraction in tagged MRI using generalized information measures.Medical Image Analysis, 9(4), 353-375.

    [7] Cocosco, C. A., Netsch, T., Sénégas, J., Bystrov, D., Niessen, W. J., and Viergever, M. A. (2004, June). Automatic cardiac region-of-interest computation in cine 3D structural MRI. In International Congress Series,1268, 1126-1131.

    [8] Huang, J., Huang, X., Metaxas, D., and Axel, L. (2007, April). Dynamic texture based heart localization and segmentation in 4-d cardiac images. In Biomedical Imaging: From Nano to Macro(ISBI), 852-855.

     

     

    [9] Jolly, M. P. (2008). Automatic recovery of the left ventricular blood pool in cardiac cine MR images. In Medical Image Computing and Computer-Assisted Intervention–(MICCAI), 110-118.

    [10] Pednekar, A., Kurkure, U., Muthupillai, R., Flamm, S., and Kakadiaris, I. A. (2006). Automated left ventricular segmentation in cardiac MRI. Biomedical Engineering, IEEE Transactions on, 53(7), 1425-1428.

    [11] Gering, D. T. (2003). Automatic segmentation of cardiac MRI. In Medical Image Computing and Computer-Assisted Intervention-MICCAI, 524-532.

    [12] Lin, X., Cowan, B. R., and Young, A. A. (2006). Automated detection of left ventricle in 4D MR images: experience from a large study. In Medical Image Computing and Computer-Assisted Intervention–MICCAI, 728-735.

    [13] Jolly, M. P., Duta, N., and Funka-Lea, G. (2001). Segmentation of the left ventricle in cardiac MR images. In Computer Vision, 2001. ICCV 2001, 1, 501-508.

    [14] Viola, P., and Jones, M. (2001). Rapid object detection using a boosted cascade of simple features. In Computer Vision and Pattern Recognition, 2001. CVPR 2001, 1, 511-518.

    [15] Goshtasby, A., and Turner, D. A. (1995). Segmentation of cardiac cine MR images for extraction of right and left ventricular chambers. Medical Imaging, IEEE Transactions on, 14(1), 56-64.

    [16] Weng, J., Singh, A., and Chiu, M. Y. (1997). Learning-based ventricle detection from cardiac MR and CT images. Medical Imaging, IEEE Transactions on, 16(4), 378-391.

    [17] Katouzian, A., Prakash, A., and Konofagou, E. (2006, August). A new automated technique for left- and right-ventricular segmentation in magnetic resonance imaging. In Engineering in Medicine and Biology Society, 2006. EMBS'06. 28th Annual International Conference of the IEEE, 3074-3077.

    [18] Nachtomy, E., Cooperstein, R., Vaturi, M., Bosak, E., Vered, Z., and Akselrod, S. (1998). Automatic assessment of cardiac function from short-axis MRI: procedure and clinical evaluation. Magnetic resonance imaging, 16(4), 365-376.

    [19] Gupta, A., Von Kurowski, L., Singh, A., Geiger, D., Liang, C. C., Chiu, M. Y., and Wilson, D. L. (1993, September). Cardiac MR image segmentation using deformable models. In Computers in Cardiology 1993, Proceedings, 747-750.

    [20] Uzümcü, M., van der Geest, R. J., Swingen, C., Reiber, J. H., and Lelieveldt, B. P. (2006). Time continuous tracking and segmentation of cardiovascular magnetic resonance images using multidimensional dynamic programming. Investigative radiology, 41(1), 52-62.

    [21] Van der Geest, R. J., Jansen, E., Buller, V. G. M., and Reiber, J. H. C. (1994, September). Automated detection of left ventricular epi- and endocardial contours in short-axis MR images. In Computers in Cardiology, USA, 33-36.

    [22] Yeh, J. Y., Fu, J. C., Wu, C. C., Lin, H. M., and Chai, J. W. (2005). Myocardial border detection by branch-and-bound dynamic programming in magnetic resonance images. Computer methods and programs in biomedicine, 79(1), 19-29.

    [23] Liu, N., Crozier, S., Wilson, S., Liu, F., Appleton, B., Trakic, A., and Riley, R. (2006, January). Right ventricle extraction by low level and model-based algorithm. In Engineering in Medicine and Biology Society, IEEE-EMBS, 1607-1610.

    [24] Geiger, D., Gupta, A., Costa, L. A., and Vlontzos, J. (1995).

Presenting a method for right and left ventricle segmentation from cardiac MRI images