«Chair for Computer-Aided Medical Procedures & Augmented Reality Improvement and Automatic Classification of IVUS-VH (Intravascular Ultrasound – ...»
TECHNISCHE UNIVERSITÄT MÜNCHEN
Chair for Computer-Aided Medical Procedures & Augmented Reality
Improvement and Automatic Classification of
IVUS-VH (Intravascular Ultrasound – Virtual Histology)
Vollständiger Abdruck der von der Fakultät für Informatik der Technischen
Universität München zur Erlangung des akademischen Grades eines
Doktors der Naturwissenschaften (Dr. rer. nat.)
Vorsitzende: Univ.-Prof. Gudrun J. Klinker, Ph.D
Prüfer der Dissertation:
1. Univ.-Prof. Nassir Navab, Ph.D.
2. Assoc. Prof. S. Kamaledin Setarehdan, Ph.D.
University of Tehran, Iran
3. Asst. Prof. Gozde Unal, Ph.D.
Sabanci University Istanbul, Türkei Die Dissertation wurde am 17.02.2010 bei der Technischen Universität München eingereicht und durch die Fakultät für Informatik am 18.08.2010 angenommen.
Acknowledgments I would like to express my sincere gratitude to my PhD adviser Professor Dr. Navab for his guidance, support and encouragement during this research, as well as my entire period of studies at Technical University of Munich.
I am deeply grateful to Dr. Unal and Dr. Setarehdan, my supervisor at university of Tehran, for their invaluable support, creative ideas, detailed and constructive comments, throughout this research. The latter has been a source of inspiration throughout my postgraduate studies.
¨ My thanks goes to Dr. Rieber and most importantly Dr. Konig, who were the motivating factor in choosing my subject matter and had a remarkable inﬂuence in this work; their generosity to share their data will long be remembered.
During this work I have collaborated with many colleagues, for whom I have deepest respect and, therefore, wish to extend my warmest thanks to Reza Mousavi, Olivier Pauly, Amin Katouzian, Ali Bigdelou and Holger Heterrich who all helped with their detailed reviews, constructive criticism and fruitful discussions.
Special thanks go to Sara Avansari and Alireza Roodaki for their sympathetic help and valuable input in this work.
Furthermore, I would like to thank Martina Hilla, for all her cooperation during my PhD program.
I would also like to thank to my sister and her husband, Elham and Dr Alireza Omrani, and specially my brother, Salar, who have supported me during my years of studies.
Last, but not least, I owe especial thanks to my parents for their patience and support during this thesis. Their understanding and support has been exceptional and I feel forever indebted.
Without any of the above this thesis would not have been completed.
Heart attack and stroke are the major causes of human death and atherosclerotic plaques are the most common effect of cardiovascular disease. Intravascular ultrasound (IVUS), a diagnostic imaging technique, offers a unique view of the morphology of the arterial plaque and displays the morphological and histological properties of a cross-section of the vessel. Limitations of the grayscale IVUS manual plaque assessment have led to the development of quantitative techniques for analysis of characteristics of plaque components.
In vivo plaque characterization with the so called Virtual Histology (VH)-IVUS, which is based on the ultrasound RF signal processing, is widely available for atherosclerosis plaque characterization in IVUS images. However, it suffers from a poor longitudinal resolution due to the ECG-gated acquisition. The focus of this PhD thesis is to provide effective methods for image-based vessel plaque characterization via IVUS image analysis to overcome the limitations of current techniques. The proposed algorithms are also applicable to the large amount of the IVUS image sequences obtained from patients in the past, where there is no access to the corresponding radio frequency(RF) data. Since the proposed method is applicable to all IVUS frames of the heart cycle, it outperforms the longitudinal resolution of the so called VH method.
The procedures of analyzing grayscale IVUS images can be divided into two separated aspects: (i) detecting the vessel borders to extract the region called ”plaque area”. (ii) characterizing the atherosclerosis plaque composition. The latter one consists of two main steps: in the ﬁrst one, known as feature extraction, the plaque area of the cross-sectional IVUS image is modeled using appropriate features. The second step based on learning techniques assists the classiﬁer in distinguishing different classes more precisely and in assigning labels to each of the samples generated by feature extraction within the ﬁrst step.
In-vivo and ex-vivo validation procedures were used, where the results proved the efﬁciency of the proposed algorithms for vessel plaque characterization via IVUS images. A graphic user interface (GUI) is designed as an effective image processing tool which enables cardiologists with a complete IVUS image processing tool from border detection to plaque characterization. The algorithms developed within this thesis leads to the enhancement of the longitudinal resolution of plaque composition analysis. In the ﬁnal part of the thesis, this is shown analytically and is highlighted by presenting a three dimensional view of both the vessel and the distribution of different plaque components in the plaque area.
IVUS, Plaque Component Analysis, Longitudinal Resolution, Virtual Histology
CHAPTER 1: MEDICAL BACKGROUNDAn introduction to medical basics is focused in this work and the clinical applications of several medical imaging modalities, esp. IVUS and Virtual Histology, used for atherosclerotic plaque assessment followed by a description of problem statement.
CHAPTER 2: TECHNICAL PRINCIPLES OF IVUSA general review of ultrasound physics and technical principals of IVUS imaging.
CHAPTER 3: PLAQUE AREA DETECTIONA review of existing techniques for border detection and an introduction to algorithms employed in this work.
CHAPTER 4: STATE OF THE ART IN PLAQUE CHARACTERIZATIONA review of previous techniques used for IVUS plaque characterization.
CHAPTER 5: NEW APPROACHES FOR PLAQUE COMPONENT ANALYSISFour new algorithms presented based on image texture analysis methods for characterization of coronary plaque composition via IVUS images.
CHAPTER 6: VALIDATIONImplementing in-vivo and ex-vivo validation procedures and also presenting the enhancement of the longitudinal resolution of plaque composition analysis.
CHAPTER 7: CONCLUSION AND FUTURE WORKA short summary followed by a discussion of four proposed algorithms for plaque analysis and future works.
A. LIST OF PUBLICATIONSAll publications contributed to the scientiﬁc community during this work.
B. LIST OF ABBREVIATIONS
C. QUANTITATIVE ANALYSIS FOR CLINICAL REPORTS OF IVUS IMAGING
1.1. Biological Concerns Despite signiﬁcant advances in diagnosis and treatment, coronary atherosclerosis remains a major cause of death in developed countries  and almost twice as many people die from cardiovascular diseases than from all forms of cancer combined.
1.1.1. Biological Background The coronary arteries are the vessels which supply the heart muscle with blood and therefore oxygen. The right coronary artery and the left main coronary artery are the two branches that come from the aorta to feed the back of the heart and its front respectively (ﬁgure 1.1). The left main coronary artery further splits up in the left anterior descending(LCD) and the left circumﬂex(LCX), to supply with blood the front of the heart and the left side and back of the heart .
Figure 1.1.: Coronary arteries .
These coronary arteries mainly consist of three layers. The inner layer is intima, the middle layer is media, and the outer one is adventitia. These layers are observable in a cross-section of the artery (ﬁgure 1.2). Media consists of smooth muscle cells, with an approximate thickness of 100 microns. The adventitia is made up of mostly elastic and collagen ﬁbers and ﬁbroblasts. The intima has a variable thickness which is expressed as a ratio of media thickness. Ratios of 0.1 to 1 are considered normal .
1.1.2. Coronary Artery Disease Atherosclerotic plaques are the most common effect of cardiovascular disease. Atherosclerotic plaques can either cause stable coronary artery stenosis leading to angina pectoris
1. Chapter One:
Medical Background Figure 1.2.: Cross-section of a coronary artery which shows intima, media, and adventitia layers .
during exercise, or lead to acute coronary or vascular events such as myocardial infarction or stroke when they rupture.
The development of atherosclerotic lesions can be categorized in six classes based on the deﬁnition of the Committee on Vascular Lesions of the Council on Atherosclerosis of the American Heart Association. Classes I-III are considered precursors of an established lesion and had already been seen in children and adolescents. Types IV-VI are classiﬁed as advanced atherosclerotic lesions .
Atherosclerosis takes place due to an activated endothelium, mainly in areas with preexisting intima thickening. An activated endothelium is characterized by a raised adhesiveness for monocytes due to over-expression of adhesion molecules, an enhanced permeability to lipoproteins or functional imbalances of pro- and anti-thrombotic factors, growth stimulators and inhibitors, and vasoactive substances.
1.1.3. Lesions Associated with Coronary Artery Disease There are different types of plaques responsible for causing coronary atherosclerosis. For years, rupture of vulnerable plaques was addressed as the most important sources of coronary thrombosis. But other types of plaques and lesions also exist in the literature .
Three different plaque etiologies associated with coronary thrombosis is shown in table
1.1. In the following sections these lesions will be discussed in detail.
Rupture of Vulnerable Plaque The rupture of vulnerable plaques is the main cause of acute coronary syndrome , . Acute Coronary Syndrome (ACS) (or Myocardial Infarction (AMI) /unstable angina) occurs when the supplied oxygen from the coronary arteries is less than the myocardial demand. It usually happens due to atherosclerotic coronary artery disease (CAD). In this case, when the atheromatous plaque builds up on the wall of the coronary arteries, the lumen of the artery will be probably compromised.
Stenotic plaques which compromise the coronary arteries luminal surface area by more than 60-70 percent used to be clinically signiﬁcant and hence, a patient with this condition used to be marked as high risk for ACS. However, many researches have shown that in some cases the AMI occurs due to the occlusion of coronary arteries without any significant stenosis. It means sometimes plaques causing severe stenosis are not the cause of
The plaque at a high risk of rupturing was suggested by the researchers in place of stenosis plaque as the immediate precursor of most of the culprit plaques for ACS stenotic .
This, indeed, was the ﬁrst step in the development of the concept of Vulnerable Plaque (VP). The VP is the ”short-term precursor” to the culprit plaque, which triggers clinical ACS.
Although the lipid core encased in ﬁbrous ”cap” plaques may compromise the lumen of the coronary, based on histopathological studies of ”culprit” plaques, approximately 60 to 70 percent of AMIs are caused by plaque rupture, with release of the thrombogenic core of lipid and necrotic debris , . Based upon these ﬁndings, VPs are deﬁned as plaques at high risk for rupture, or for having the surface of their ﬁbrous cap denuded in either case leading to thrombus formation. This may weaken the routine method of clinical decision-making based on determining the stenosis because it would be unreliable due to an ignoring of the major cause of AMI. VPs are supposed to have three histologic hallmarks compared to stable ones: a larger lipid core, a thinner ﬁbrous, and many inﬂammatory cells.
In addition to the above deﬁnition, VP can be described as any plaque that might cause clinically signiﬁcant CAD. In this deﬁnition, any plaque vulnerable to rupture or denudation is deﬁned as VP. While covering the previous deﬁnition of VP, many other plaques are
1. Chapter One:
Medical Background marked as VP (ﬁgure 1.4).
Plaque A is the main cause of culprit lesions and represents the previous deﬁnition of VP . To complete the deﬁnition, VP criteria are expanded as follows: Major criteria include: active inﬂammation (monocyte/macrophage +/- Tcell inﬁltration), thin cap with large lipid core, endothelial denudation with superﬁcial platelet aggregation, ﬁssured plaque, and more than 90 percent stenosis. Minor criteria include: superﬁcial calciﬁed nodule, glistening yellow appearance, intraplaque hemorrhage, endothelial dysfunction, and outward (positive) remodeling.
Plaque Erosion: Plaque erosion is referred when a thrombosed arterial segment fails to show an area of rupture after serial sectioning . The exposed intima at the plaque thrombus interface consists predominantly of smooth muscle within a proteoglycan-rich matrix with minimal inﬂammation. Typically, the endothelium is absent at the erosion site.
The development of erosion may be the sequel to repeated episodes of focal vasospasm;
however, to date, there has been no direct evidence to support this hypothesis.