MANAV RACHNA INTERNATIONAL INSTITUTE OF RESEARCH AND STUDIES

(DEEMED TO BE UNIVERSITY)

Short Synopsis

for

Ph. D. Programme 2015-16

Development of an Algorithm for Object Detection in Medical Images using Image Segmentation and Deep Learning Techniques

DEPARTMENT OF COMPUTER APPLICATIONS

FACULTY OF COMPUTER APPLICATIONS

Submitted by:

Research scholar:

Shiv Shankar RaiRegistration No.: 15/Ph.D./008Email ID: shivshn33@gmail.comContact No.: 8826304963

Supervisor:

Dr. (Mrs) Shaveta BhatiaAssociate ProfessorEmail ID: shaveta.fca@mriu.edu.inContact No.: 8800182889

``1

ABSTRACT

Heart attack is mainly caused due to atherosclerosis. It is a coronary artery disease (CAD) and it is a leading cause of death worldwide. It occurs when the coronary artery that supplies blood and oxygen to the heart and different parts of the body becomes blocked or narrowed due to deposition of proteins, cholesterol and other fatty deposits in the inner wall of the coronary artery. This results in a heart attack or damage to the heart tissue. Existing techniques for detection of plaque are Magnetic Resonance Imaging (MRI), Electron Beam Computed Tomography (EBCT) and Angiography, among these existing techniques Angiography is widely used at present to detect and cure heart attack. The currently available techniques are expensive, beyond the reach of normal people, not easily available in remote areas and cannot detect deeply embedded plaque with accuracy at early stages. The plaque is only detected by these techniques when blockage in artery is more than 70%.In order to remove the limitations of existing techniques Intravascular Ultrasound (IVUS) technique is proposed in our research work. This technique is based on application of image processing (MATLAB SOFTWARE) by using bio-markers and Marker Controlled Watershed Modified Image Segmentation Algorithm which better detects the Plaque deposits embedded deep inside the Coronary Artery wall that are undetectable by Angiography or other Cardiac test that are prone to rupture without warning sign. In addition to this technique Artificial Intelligence (Deep Learning) computing technique will also be used for image classification. The proposed technique has tendency to detect inner layer (foreground) i.e. object and outer layer (background) of the artery for better detection of Region of Interest (ROI), which is deeper region of soft non-calcified plaque. So, the result obtained using the proposed techniques is expected to play a crucial role in the visualization of inner view of Coronary Artery containing Plaque with more accuracy and also expected to show the exact measurement of plaque location, size and quantification for pre-detection of heart attack.

Keywords: Atherosclerosis, Intravascular Ultrasound (IVUS), Watershed Algorithm, Deep

Learning and Region of interest (ROI).

``2

CONTENTS

S. No. Description Page No.

1 Introduction 4

2 Literature Review 4-8

3 Description of Broad Area 8

3.1 Current Techniques 8

3. 2. (A). Intravascular Ultrasonography (IVUS) Digital image processing technique: 8

3.2. (B). Image classification using deep learning technique of artificial intelligence 9

4 Motivation of the Research 9

5 Objectives of the Research 10

6 Methodology to be adopted 10

7 Proposed Algorithm 11

8 Expected Outcome of the Research 12

9 References 13-15

``3

1INTRODUCTION

The Coronary Artery Disease (CAD) or Ischemetic Heart Disease (IHD) is caused due to the Plaque formation in an inner artery wall. The deposition of Plaque or Atherosclerosis is a reason for stroke or heart attack. Plaque is a blockage or deposits of proteins, cholesterol and other fatty substances that blocks the flow of blood and oxygen to the heart and causes either damage to heart tissues or leads to heart attack. The proposed research work is concerned with the pre- detection of stroke or heart attack. The proposed work started with the study and analysis of Angiography, Intravascular Ultrasound (IVUS) Images and other techniques for detection of plaque in Coronary Artery. After study, it was analysed that there are many limitations in the existing techniques, but the Intravascular Ultrasound (IVUS) technique is better among available techniques therefore in the proposed research work for pre-detection of stroke or heart attack and to improve the effectiveness and efficiency of diagnostic process of Coronary Artery Disease (CAD) , the Intravascular Ultrasound (IVUS) technique, along with Deep Learning technique of Artificial Intelligence will be used. The Intravascular Ultrasound (IVUS) is used for scanning and recording of moving structures. The software namely “DICOM VIEWER” is used for converting scanned video files into multiple images. The MATLAB digital image processing software will be applied on images for image segmentation. The new algorithm is proposed as an enhancement over Marker Controlled Watershed Modified Image Segmentation algorithm which further consists of image processing and Image segmentation process. The steps of Image pre-processing are:

1. Setting of image attributes2. Conversion of image from RGB to GRAY scale 3. Removal of Speckles or Noise from image

After Image pre-processing Image segmentation will be applied. The steps of Image segmentation are:1. Computation of Gradient Magnitude 2. Computation of Watershed Transform3. Computation of Internal Region 4. Computation of External Region

At last, Image Classification will be done using Deep Learning technique. The steps of Image Classification are:

1. Image Convolution2. Image Pooling3. Image Concatenation4. Image Sampling

2. LITERATURE REVIEW

Anju Bala [1], has contributed a noble method called Improved Watershed Image Segmentation Technique using MATLAB which is applicable to find the catchment basins to imagine the landscape being immersed in a lake in current work prewitt‘s operator was used to detect image edges rather than sobel operator to avoid over segmentation.

The advantages of this method over the other existing method are that in practice, the watershed is often applied to the image gradient in order to separate the homogenous region which gives the desired result. The disadvantages of this process are the quality of the gradient which has a major influence on the performance of the segmentation process.

``4

Thodeti et al [11], has presented a method for colour image segmentation using watershed algorithm. This method shows the colour and texture that are the two important features of information retrieval in images and video, colour image segmentation is informative although it has various limitations. The gray scale image segmentation methods are not applicable to the colour images which will be carried out in the proposed approach.

The advantages of this method is that it is widely used almost in every field that is related to image processing like remote sensing image, medical image and traffic image. The researchers used this smethod for framing and developing classical image segmentation algorithm like threshold based and area based image segmentation algorithm to detect the edges in an image.

The disadvantages of this type of method application i.e. traditional algorithms are that they may be confronting with feature extraction problem, uncertainty and fuzziness in an image which reduces the image segmentation result.

Faouzi Benzarti and Hamid Amiri [3], has presented an image denoising approach using diffusion tensor and homographic transformation to adapt the flow of diffusion by applying anisotropic diffusion along the coherent structure direction of the echo graphic images which is interesting features in 3D Ultrasound images. This approach allows diffusion with the condition of additive noise in 3D image denoising.

The disdvantages of this method is that it adapts the flow of diffusion towards the local orientation by applying aniostropic diffusion along the coherent structure dissection of the image interesting feature or areas of interest.The disadvantages of this method are it does not work for 3D ultrasound image denoising.

R.Sukanesh et.al [8] has developed a technique for image noise reduction by Wavelet Thresholding based on weighted variance using standard speckle filters‖. The new version used here was an adaptive filter wiener 2 in the MATLAB image processing toolbox and signal to noise ratio (SNR) of each denoised image was also calculated.

The advantages of this method is that in ultrasound images, the noise can restrain information which is valuable for the general practioner, medical images are very inconsistent and need to study case by case which has been done by them in their current work.

The disadvantage of this method is that, they have used adaptive filter which generally incorporates other filters like kuan filter.

Firas Wada and Mohamed Ali Hamdi[7] has developed a method called 3D Segmentation of Intravascular ultrasound Images by using Fast Marching method and the goal of this method is to demonstrate the IVUS segmentation using the method of curvelet to improve the IVUS image quality.

The disadvantages of this method is that they have prepared the method of curvelet which is capable of improving IVUS images original quality using Rayleigh Probability Density Function (PDF), mixture and realistic simulation of IVUS images.

The disadvantage of this method is that the ivus image front does not explore its initial within the region.

``5

S.Satheesh and Dr.KVSVR Prasad [10] has proposed a method for Gaussian noise removal from MR IMAGES of biomedical using algorithm which is efficient and is based on the contour let transform for Gaussian noise for both the qualitative and quantitative analysis.

The advantages of this method is that they have converted the speckle image into a transform domain like wavelet and counterlet for comparing the coefficient with fixed threshold and they have designed the new algorithm which defines the new threshold value to eliminate the corrupted pixels of an image

The disadvantages of the method is that the algorithms works not only for the removal of Gaussian noise alone for qualitative and quantitative analysis but different images may contain several other types of noises for which suitable noise removal techniques and algorithm has to be designed and applied for better output or results

Kalpana saini et al [8] has proposed the idea of image segmentation in ultrasound images for diagnosis and therapeutic purposes which shows that in order to enhance the feature of an image which requires further pre-processing there are various techniques out of which the image segmentation is expected to give better results in comparison to other techniques and it is expected that even for detection of plaque in an IVUS image it will be beneficial.

Anthony N. De Mari [6] has proposed comparative technique for vulnerable plaque in this comparative work, it was suggested that IVUS has the ability not only to characterize plaque either as calcified or fibro-fatty but could not detect lipid-rich plaque, necrotic core and thrombus but according to the recent advances in medical application of IVUS, it is possible to detect lipid-rich plaque by back scattered wavelet analysis.

The advantages of this method is that using the integrated backscattered, wavelet analysis and virtual histology, this method has ability to evaluate and mathematically transform the radiofrequency signal into a color-coded representation from ultrasound waves in order to characterize lipid, fibrous tissues, calcification and necrotic core. The aim of this method is plaque component highlighting. This technique is advantageous for vulnerable plaque detection.

The disadvantages of this method is that results shows that plaque observed using this method are concentric, positively modeled and relatively free of calcification and have 3D reconstruction which allows morphological assessment of lesions, but is not useful in detecting clinically relevalent plaque characteristics such as lipid.

Alka Vishwa and Shilpa Sharma [5] they have developed a wavelet-based thresholding scheme for noise removal from IVUS images. In their work, they have used adaptive and anisotropic diffusion technique to remove noise from different types of images like ultrasonography images and they have obtained better results. The same results may be possible by analyzing carotinoid artery scan images.

``6

The advantages of this method is that they have used various filters in order to remove noise from image for better vision of plaque and also improve the diagnostic examination to increase the efficiency of segmentation technique.

The disadvantages of this technique is that different images taken at different degree of deformation may contain different type of noises for which suitable filter has to be designed which may differ from one image to the other image for better object detection in an image vision.

Ahmad EL ALLAOUI and M‘barek NASRI [3], has developed an approach for medical image segmentation in order to avoid the problem of over-segmentation which is a major problem of watershed transform .The key points in this approach are morphological reconstruction,, extraction of markers and application of watershed transform which is a line of separation in an image of greatest intensity. Felix Renard and Yongyi Yang [12], has contributed a noble method called, Image analysis for detection of coronary artery soft plaques in MDCT images using image segmentation which is applicable to segment three region i.e. lumen of artery extraction, centraline points vessel direction and arterial wall.

Harvey S. Hecht, MD, FACC, H. Robert Superko. et. al. [23] , has proposed a method called Coronary calcium as a risk factor: role in global risk assessment in which they have determined the relationship between national Cholesterol Education Program (NCEP), ATP-II guidelines and clinical atherosclerosis for female patient using electron beam computed tomography.

Stephan Achenbach,MD Fabian Moselewski. et.al. [20], has proposed the technique for Detection of Coronary Atherosclerotic Plaque by Contrast-Enhanced for both calcified as well as non-calcified plaque region by Sub millimeter. The Multi detector Spiral computed tomography in Comparison to Intravascular Ultrasound is not accurate.

Udo Hoffmann, Maros Ferencik et.al.[11], has proposed Coronary CT Angiography for noninvasive detection of coronary artery atherosclerosis by multi detector tomography for children and adolescent using risk prediction algorithm can also reduces the effect of hypertension and hyperlipidemia. In their current work they have also appreciated the upcoming technique like ultrasonography which is a proposed technique in this paper.

NCEP- II. Expert Panel on Detection, Evaluation and Treatment of Cholesterol of blood in Adults‖[24], is the report of the NCEP, Expert Panel on Detection, Evaluation, and High Blood Treatment of Cholesterol in Adults for better analysis and detection of plaque.

Leber AW, White CW, et al. [22], has contributed a noble method called Composition of coronary plaques having acute myocardial infarction in patients with stable angina pectoris determined by CT‖ in which the technique is used to search the burdens and composition of plaque analysis and determination of contrast-enhanced multi-sliceButler J, Shapiro M et.al [16], has proposed a method called Computed Tomography‖, the current work is a technique to differentiate and categorize the non-invasive CAD to improve the medical therapy.

``7

Corbin E. Goerlich, O. H. Frazier & William E. Cohn [13], has presented a method to improve previous challenges and current progress–the use of total artificial hearts patients with last heart failure‖, it is a Review of Cardiovascular Therapy by experts, in which they have proposed a method to make the transplant of heart i.e. LVAD a self-contained bio compatible heart (TAH) called total artificial.

Achenbach S, Moselewski F[17], has proposed a technique for detection of calcified and non-calcified coronary atherosclerotic plaque by multi-detector spiral computed tomography: a segment-based method of IVUS‖, in which they have presented the idea of quantification of calcified and non-calcified plaque deposits in less than or equal to 3 mm in diameter which are difficult to quantify.

3. DESCRIPTION OF BROAD AREA

The main aim of the proposed research work is pre-detection of plaque; the proposed research work is based on the Intravascular Ultrasound (IVUS) technique. This technique consists of recording of internal regions of Coronary Artery. Further these recording regions or moving pictures will be converted into multiple images using DICOM Viewer for detection of objects in an images in addition to two techniques namely Image segmentation using Marker Controlled Watershed Modified algorithm of MATLAB software and Deep Learning technique of Artificial Intelligence will be applied to achieve the expected result.

3.1. Current Techniques:

There are three existing techniques that are used to analyze and detect Coronary Artery Plaque (CAD) by Physicians and Cardiologists, current techniques have various drawbacks like they cannot detect the deeply embedded soft plaque at early stages. Current techniques are invasive in nature and their operating expenses are higher in terms of cost and are beyond the reach of normal people. The description of current techniques is as follows:

(A). Angiography: It is a medical imaging technique widely used at present to analyze and detect coronary artery plaque. This technique is invasive and is performed using a gauge wire inserted into the heart through arms tissues to analyze, detect and visualize the internal plaque deposits that causes attack. The complications of X-ray angiography test is limited by the invasive nature of the procedure and by the use of potentially harmful radiation and contrast dye that is major drawback of this technique. This technique cannot detect the soft plaque deeply embedded at early stages that further leads to severe attack.

(B). Magnetic Resonance imaging (MRI): It is a medical imaging technique that uses magnetic field and radio waves to create detailed images of the coronary artery. Radio waves cause the aligned atoms present in radio waves to produce very faint signals, which are used to create cross-sectional MRI images for plaque detection. This technique does not detect the deeply embedded soft plaque deposits in the coronary artery.

``8

(C). Electron Beam Computed Tomography (EBCT): is a specific form of computed tomography (CT) in which the different designs have been developed explicitly for better images scanning of an artery that performs a complete cycle of movement with each heart beat. This is a non-invasive technique which is used to detect the calcium deposits in the interior artery wall that causes blockage and atherosclerosis inside the Coronary Artery. The limitation of this technique is it cannot detect the fatty substances deposition at early stages.

3.2. Proposed Techniques:

The proposed work will be implemented using Intravascular Ultrasound (IVUS) Image Segmentation Technique of Image Processing using Matrix laboratory (MATLAB) Software and Deep Learning technique of Artificial Intelligence (AI). The description of both the techniques are: Image Segmentation technique: It is a medical Image Segmentation technique in which initial step is analysis and recording of heart scan, where video scans are converted into multiple images using DICOM viewer. The medical images are usually corrupted with noises called speckle that hides the necessary information of an image In order to remove these noises image needs to be pre-processed. The Pre-processing steps include setting of image attributes, conversion of image from RGB to GRAY scale and removal of noises if any. Image segmentation is next step to image processing that includes application of watershed using bio-markers in order to distinguish between inner and outer layer for visualization of region of interest (ROI).

(B). Deep Learning technique of Artificial Intelligence (AI):

Deep learning technique of Artificial Intelligence will be used for detection and classification of objects and regions using image convolution at different rates by sampling and image pooling. It is used to visualize the exact location of the region of interest (ROI), i.e. plaque deposition in an inner coronary artery for pre detection of plaque.

4. MOTIVATION OF THE RESEARCH

The motivation of the proposed work is due to the lack of methodologies and techniques which cannot effectively and efficiently detect heart attack/stroke at earlier stage or pre detection of heart attack that leads to death of a patient and suffering of their near and dear ones. Lack of cost effective techniques, lack of easily availability in rural areas and inaccurate methods to analyze and early detection of the deadliest disease called heart attack motivated the author to continue the research work in this particular area in medical field. In order to overcome the above limitations, the new algorithm is proposed that has tendency to fulfil the needs of poor and needy patients because the proposed techniques are cheaper and can be made easily available in all the labs to provide better treatment to the patients. It will be also applicable for different purposes wherever it is needed to detect the objects and regions in different set of data in medical images.

``9

5. OBJECTIVES OF THE RESEARCH

1. To review literature related to object detection in medical images /videos for various parameters

2. To develop an algorithm as an enhancement of Marker Controlled Watershed Image Segmentation algorithm

3. To detect and classify objects and regions in an image using Deep Learning technique4. To implement the proposed algorithms as a Graphical User Interface (GUI) based tool in

MATLAB.

Plan of research

Collect the Images of Intravascular Ultrasound Scan of heart attack patients which are recorded using IVUS technique and some other medical images where image segmentation is applicable to detect the region of interest.

1. Apply image pre-processing steps of digital imaging (DI) like setting of image attributes, image conversion from colour image to gray scale and noise reduction.

Apply new proposed algorithm which is an enhancement over Marker Controlled Watershed Algorithm to detect plaque.

Detect and classify the objects and regions using Deep Learning Algorithm of Artificial Intelligence (AI) for image prediction and image classification.

Compare the results obtained using proposed algorithm with other existing algorithms for checking the accuracy of results.

6. METHODOLOGY TO BE ADOPTED

A real global application of image processing involves medical images and image segmentation technique that has been developed for different applications. Blockage in an artery is a condition in which heart does not receive sufficient amount of blood and oxygen and causes heart attack. Global statistics shows that attack is now a day’s common in children and adolescents and it is likely to increase the number because due to lack of accurate technique to detect soft plaque deposits that are deep embedded in an inner artery wall. In order to provide valuable support to the physicians and cardiologist for detection of plaque, this algorithm is proposed by combining ultrasound and digital image segmentation techniques that uses marker controlled watershed modified image segmentation algorithm. This algorithm is based on acquiring of ultrasound image, setting of image attributes, image conversion from colour to gray scale, image enhancement, computation of watershed transform, computation of bio- markers (foreground and background), image extraction, image comparison and expected result visualization. Similarly this technique will also be applicable for different data set of a medical images where region of interest can be detected using this proposed algorithm.

7. Proposed Algorithms

``10

The proposed Algorithms are:

A). An Enhancement over Marker Controlled Watershed Image Segmentation Algorithm: The

proposed algorithm is based on image segmentation technique which is considered to be the best

algorithm for visualization of the object or plaque containing region in medical images in order to

distinguish between the inner and outer region of coronary artery. The flow chart of the Proposed

Algorithm is shown below:-

FIG: SHOWING THE STEPS OF IMAGE SEGMENTATION

B). ARTIFICIAL INTELLIGENCE (DEEP LEARNING ALGORITHM)

``11

INPUT IVUS IMAGE

CONVERT IVUS IMAGE INTO GRAY SCALE

SET IMAGE ATTRIBUTES

APPLY GRADIENT MAGNITUDE

APPLY WATERSHED TRANSFORM

ENUMERATE IVUS INTERNAL REGION

ENUMERATE IVUS EXTERNAL REGION

RESULT VISUALIZATION

IVUS IMAGE PRE PROCESSING STEPS

IVUS IMAGE SEGMENTATION STEPS

Input(ivus image)

Output

FIG: SHOWING THE DEEP LEARNING FLOWCHART OF IMAGE CLASSIFICATION

In the above figure IVUS scanned image is the input to the system which undergoes encoding and decoding steps of the image using image convolution and image pooling for sampling an image at different rate of image convolution for image segmentation and its features classification.

8. EXPECTED OUTCOME OF THE RESEARCH

The expected outcome of the research is visualization of region of interest i.e., object and regions in medical images, soft plaque that is deeply embedded in the inner artery wall that cannot be early detected by currently available techniques like angiography and others. The proposed work will distinctly visualize the artery wall and plaque deposits with better image quality that will be of great support to the physicians and cardiologist to cure and eradicate the problem of heart attack at early stage. This proposed work will detect and classify the objects and regions in an image with higher level of accuracy.

9. REFERENCES

``12

ATROUS COVOLUTION

3*3 ConvRate 6

1*1 Conv

1*1 Conv3*3ConvRate 12

3*3 ConvRate18

1*1 ConvSample by 4

POOLING

Concat3*3 Conv

*

Sample by 4

1. Bala Anju (2012) 'Improved Watershed Image Segmentation Technique using MATLAB', International Journal of Scientific & Engineering Research, 3(6), pp. 2229-5518.

2. Oliver Faust U.Rajendra Acharya et.al.(2017) ‘Computer aided diagnosis of Coronary Artery Disease, Myocardial Infarction and carotid atherosclerosis using ultrasound images: A review’ European Journal of Medical Physics,33 , pp.1-15.

3. Faouzi Benzarti and Hamid Amiri (2012) 'Speckle Noise Reduction in Medical Ultrasound Images', International journal of computer science issues, 9(2), pp. 187.

4. Ahmad EL Allaoui and M‘barek NASRI (2012) 'Medical Image Segmentation by Marker-Controlled Watershed and Mathematical Morphology', the international journal of multimedia and its application, 4(3), pp. 4301.

5. Vishwa Alka and Sharma Shilpa (2012) 'Modified Method for Denoising the Ultrasound Images by Wavelet Thresholding', I.J. Intelligent Systems and Applications, 6(2074-9058),pp. 25-30.

6. Anthony N. De Maria (2001) 'Imaging Vulnerable Plaque by Ultrasound', Journal of the American College of Cardiology, 47(8), pp. C32–C39.

7. Wada, F.I.R.A.S & ali hamdi, M.O.H.A.M.E.D. 2013. 3D Segmentation of Intravascular Ultrasound Images: A Fast-Marching Method. Journal of Radiology and Diagnostic Imaging. 47(8), pp. 29-36.

8. Saini Kalpana et al. (2010) 'Ultrasound Imaging and Image Segmentation in the area of Ultrasound a Review', International Journal of Advanced Science and Technology, 24(5),pp. 3548.

9. Sukanesh, R et al. (2009) 'Speckle Noise Reduction in Ultrasound Images by Wavelet Thresholding based on Weighted Variance', International Journal of Computer Theory and Engineering, 1(1), pp. 1793-8201.

10. Satheesh, S. and Prasad Dr.KVSVR (2011) 'Medical Image Denoising using Adaptive Threshold Based on Contourlet Transform Advanced Computing', An International Journal (ACIJ), 2(2), pp. 2206.

11. Thodeti Srikanth et al. (2011) 'Color Image Segmentation using Watershed Algorithm', International Journal of Computer Science and Information technologies (IJCSIT), 2(5), pp. 2332-2334.

12. U Hoffmann and J Butler, (2005) 'Noninvasive detection of coronary atherosclerotic plaque by multi detector row computed tomography', International Journal of Obesity, (), pp. S46–S53.

13. Felix Renard and Yongyi Yang IEEE International Symposium on Biomedical Imaging: from nano to macro. IEEE International Symposium on Biomedical Imaging June 2008 (2008) 'Image analysis for detection of coronary artery soft plaques in MDCT images', IEEE International Symposium on Biomedical Imaging: from nano to macro. IEEE International Symposium on Biomedical Imaging, 10(10), pp. 1109.

14. Corbin E. Goerlich, O. H. Frazier & William E. Cohn (2016) 'Previous challenges and current progress–the use of total artificial hearts in patients with end-stage heart failure', Expert Review of Cardiovascular Therapy, 14(12), pp. 1095-1098.

15. Udo Hoffmann, Maros Ferencik et al. (2006) ' Coronary CT Angiography ', The journal of nuclear medicine, 47(5), pp. 797-806.

``13

16. Rieber J, Meissner O et al. (2006) 'Diagnostic accuracy of optical coherence tomography and intravascular ultrasound for the detection and characterization of atherosclerotic plaque composition in ex-vivo coronary specimens', Diagnostic accuracy of optical coherence tomography and intravascular ultrasound for the detection and characterization of atherosclerotic plaque composition in ex-vivo coronary specimens: a comparison with histology. Coron Artery Dis, 5(30), pp. 425-430.

17. Butler J, Shapiro et al. (2007) 'Extent and distribution of coronary artery disease: a comparative study of invasive versus noninvasive angiography with computed angiography', Am Heart J., 53(3), pp. 378-8.

18. Achenbach S, Moselewski F. (2004) 'Detection of calcified and non-calcified coronary atherosclerotic plaque by contrast-enhanced, sub-millimeter multi-detector spiral computed tomography: a segment-based comparison with intravascular ultrasound.et. al, Department of Radiology, Massachusetts General Hospital, Boston, USA.', Pub Med US National Library of Medicine National Institutes of Health, 109(1), pp. 14-7.

19. Dragu, R. Kerner, A. Gruberg L. et al. (2008) 'Angiographically uncertain left main coronary artery narrowings: correlation with multidetector computed tomography and intravascular ultrasound', Int J Cardiovascular Imaging, 24(5), pp. 557-563.

20. Rasouli ML1, Shavelle DM et al. (2006) 'Assessment of coronary plaque morphology by contrast-enhanced computed tomographic angiography', comparison with intravascular ultrasound.‖, Los Angeles Bio Medical Research Institute at Harbor-UCLA Medical Center, Torrance, California, USA.,, 17(4), pp. 359-64.

21. Stephan Achenbach, MD, Fabian Moselewski. et al. Vol.10, no 1 November 2003. (2003) ' Detection of Calcified and Noncalcified Coronary Atherosclerotic Plaque by Contrast-Enhanced, Submillimeter Multidetector Spiral Computed Tomography A Segment-Based Comparison With Intravascular Ultrasound Department of Radiology Massachusetts General Hospital, Boston; and the Department of Internal Medicine, 10(1), pp. 14-7.

22. Taylor AJ, Merz NB, et al. (2003) 'Executive summary: can atherosclerosis imaging techniques improve the detection of patients at risk for ischemic heart disease? ', 34th Bethesda Conference, 41(11), pp. 250-254.

23. Leber AW, Knez A, White CW, et al. (2001) 'Composition of coronary atherosclerotic plaques in patients with acute myocardial infarction and stable angina pectoris determined by contrast-enhanced multi-slice computed tomography', AMJ Cardiol, 41(6), pp. 714-8.

24. Harvey S. Hecht, MD, FACC, H. Robert Superko. L et al. (2001) 'Coronary calcium as a risk factor: role in global risk assessment', Journal of the American College of Cardiology, 37(6), pp. 1506-1511.

25. NCEP-II Expert Panel (1993) 'Summary of the second report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High

``14

Blood Cholesterol in Adults (Adult Treatment Panel II), Pub Med US National Library of Medicine National Institutes of Health, 269(23), pp. 3015-232.

26. Emelia J. Benjamin, Michael J. Blaha et al. (2017) 'Heart Disease and Stroke Statistics— 2017 Update', A Report from the American Heart Association, 136(1), pp. 136:3-5.

27. Chirag Dodavenkannavar, Rajesh Thachathodiyl and Vikrant Vijan (2017) 'High Origin Causes High Hassle—Anomalous Radial Artery Origin Complicating a Radial Percutaneous Transluminal Coronary Angioplasty', World Journal of Cardiovascular Diseases, 7(6), pp. 207-212.

28. Rogers, WJ , Parmar, JP et al.(2010) ‗Magnetic Resonance Imaging of Carotid Atherosclerotic Plaque in Clinically Suspected Acute Transient Ischemic Attack and Acute Ischemic Stroke‘[ web of science], 22(20), pp.2031-2038.

29. Chen, WQ , Zhang, L et al.(2007) ‗Prediction of atherosclerotic plaque ruptures with high-frequency ultrasound imaging and serum inflammatory markers‘ ,American journal of physiology-heart and circulatory physiology[web of science],293(5), pp. H2836-H2844.

30. Bernard Chiu Micaela Egger (2008) ‗Development of 3D ultrasound techniques for carotid artery disease assessment and monitoring‘, International Journal of Computer Assisted Radiology and Surgery [Google scholar], 3(1-2), pp.1

31. Antonis I. Sakellarios, Kostas Stefanou et.al. (2012). ‗Novel methodology for 3D reconstruction of carotid arteries and plaque characterization based upon magnetic resonance imaging carotid angiography data‘,[Google scholar], 30(8), pp. 1068-1082

``15

``16

``17