Geometrical analysis of the sinoatrial and atrioventricular nodal region in ovine and human hearts

The cardiac conduction system is a specialized network of tissue throughout the heart that initiates and propagates electrical excitation signals which allow the heart to beat in a synchronous manner. Principal elements of the cardiac conduction system include the sinoatrial node, atrioventricular node, His bundle, left and right bundle branches, and Purkinje fibers. For years, surgeons and researchers have sought an effective method to localize these structures in procedures. Success and effectiveness of cardiac procedures like repair of critical congenital defects, right atrium and atrioventricular node ablation, and His bundle pacing depend on reliable localization of these structures. Inadvertent trauma to the cardiac conduction system leads to high levels of morbidity and mortality. Fiber-optic confocal microscopy has shown promise in locating the sinoatrial and atrioventricular nodes, but it has yet to be validated in larger mammalian hearts. This work aims to provide surgeons and researchers with a geometrical analysis of the sinoatrial and atrioventricular nodes in human and ovine hearts that will aid in the development and validation of localization devices. Through the analysis of hundreds of high-resolution serial histology images, we examined the depth, structure, microstructure, and adjacent environment of the sinoatrial and atrioventricular nodes. We utilized Masson's trichrome-stained sections along with segmentation and image processing programs designed to analyze these complex geometries. We found that the sinoatrial node resides on the right atrium epicardial surface with an average minimum depth of 55.6 μm, and the iv atrioventricular node resides beneath the right atrium endocardial surface with an average minimum depth of 171.4 μm. The depth of the atrioventricular node in ovine and human hearts increases logarithmically with age. We also reveal that the composition and microstructure of tissue superficial to the atrioventricular node can be used for reliable localization of deeper nodal structures.

GEOMETRICAL ANALYSIS OF THE SINOATRIAL AND ATRIOVENTRICULAR NODAL REGION IN OVINE AND HUMAN HEARTS by Jordan Kevin Johnson A thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Master of Science Department of Biomedical Engineering The University of Utah December 2019 Copyright © Jordan Kevin Johnson 2019 All Rights Reserved The University of Utah Graduate School STATEMENT OF THESIS APPROVAL The thesis of Jordan Kevin Johnson has been approved by the following supervisory committee members: , Chair Frank Sachse 03/15/2019 Date Approved , Member Robert W. Hitchcock 03/06/2019 Date Approved , Member Robert S. MacLeod 03/15/2019 Date Approved and by the Department of David W. Grainger , Chair of Biomedical Engineering and by David B. Kieda, Dean of The Graduate School. ABSTRACT The cardiac conduction system is a specialized network of tissue throughout the heart that initiates and propagates electrical excitation signals which allow the heart to beat in a synchronous manner. Principal elements of the cardiac conduction system include the sinoatrial node, atrioventricular node, His bundle, left and right bundle branches, and Purkinje fibers. For years, surgeons and researchers have sought an effective method to localize these structures in procedures. Success and effectiveness of cardiac procedures like repair of critical congenital defects, right atrium and atrioventricular node ablation, and His bundle pacing depend on reliable localization of these structures. Inadvertent trauma to the cardiac conduction system leads to high levels of morbidity and mortality. Fiber-optic confocal microscopy has shown promise in locating the sinoatrial and atrioventricular nodes, but it has yet to be validated in larger mammalian hearts. This work aims to provide surgeons and researchers with a geometrical analysis of the sinoatrial and atrioventricular nodes in human and ovine hearts that will aid in the development and validation of localization devices. Through the analysis of hundreds of high-resolution serial histology images, we examined the depth, structure, microstructure, and adjacent environment of the sinoatrial and atrioventricular nodes. We utilized Masson's trichromestained sections along with segmentation and image processing programs designed to analyze these complex geometries. We found that the sinoatrial node resides on the right atrium epicardial surface with an average minimum depth of 55.6 µm, and the atrioventricular node resides beneath the right atrium endocardial surface with an average minimum depth of 171.4 µm. The depth of the atrioventricular node in ovine and human hearts increases logarithmically with age. We also reveal that the composition and microstructure of tissue superficial to the atrioventricular node can be used for reliable localization of deeper nodal structures. iv TABLE OF CONTENTS ABSTRACT....................................................................................................................... iii LIST OF FIGURES ........................................................................................................... vi ACKNOWLEDGMENTS ................................................................................................ vii Chapters 1 INTRODUCTION .......................................................................................................... 1 The Cardiac Conduction System ............................................................................... 1 Intraoperative Localization of the Cardiac Conduction System ................................ 4 Fiber-Optic Confocal Microscopy ............................................................................. 5 New Ovine and Human Studies ................................................................................. 6 2 GEOMETRICAL ANALYSIS OF THE SINOATRIAL AND ATRIOVENTRICULAR NODAL REGION IN OVINE AND HUMAN......................... 8 Introduction ................................................................................................................ 8 Results...................................................................................................................... 11 Discussion ................................................................................................................ 15 Methods ................................................................................................................... 19 3 CONCLUSIONS........................................................................................................... 33 REFERENCES ................................................................................................................. 37 LIST OF FIGURES Figures 1: Sectioning and Masson’s trichrome imaging of SAN and AVN. ................................. 25 2: Analyses of Masson’s trichrome images. ..................................................................... 26 3: Masson’s trichrome images and depth analyses from 4 different anatomical regions of the AVN-His bundle complex. ......................................................................................... 27 4: Depth analyses of ovine AVN. ..................................................................................... 28 5: Coverage analyses of ovine AVN. ................................................................................ 29 6: Depth and coverage analyses of ovine SAN. ................................................................ 30 7: Example Masson’s trichrome sections of ovine AVN and SAN. ................................. 31 8: Example Masson’s trichrome sections and depth analyses of AVN and SAN in human neonatal heart. ................................................................................................................... 32 ACKNOWLEDGMENTS This work was made possible by the mentorship, help, and vision of Drs. Frank Sachse, Robert Hitchcock, and Aditya Kaza. I want to thank them for their endless patience and support. I want to recognize Azmi Ahmad and Dr. Chao Huang for giving me an amazing jump start in research and on this project. Thank you to Mahtab Waseem, Nate Knighton, Dr. Abhijit Mondal, and Bailey Kelson for being crucial members of my research team. I also acknowledge the help and time spent by my committee member Dr. Rob MacLeod in helping me through the graduation process. Finally, I wish to acknowledge the sacrifice and support of my wife Ashley and sons August and Talmage. Their love and support allowed me to do this great work. CHAPTER 1 INTRODUCTION The Cardiac Conduction System In mammals including human, the cardiac conduction system (CCS) is the network of specialized tissues throughout the heart that initiates and propagates electrical excitation signals which allow the heart to beat in a synchronous manner [1]. The effectiveness of the heart in sustaining blood flow in a body depends on a controlled, synchronized pattern of contraction and relaxation. Elements of the CCS have a wide range of functions and structures, but each element is crucial to a physiological cardiac output [2, 3]. The CCS consists of the sinoatrial node (SAN), atrioventricular node (AVN), His bundle (HB), left and right penetrating bundle branches (LBB, RBB), and Purkinje fibers (PF) [4]. The SAN is a sophisticated structure that consists of pacemaker cells organized into coiled myofibrils that reside in a highly innervated region of the heart. The SAN’s pacemaker automaticity stems from cells that express hyperpolarization-activated cyclic nucleotide-gated (HCN) transmembrane channels [5, 6]. These channels are responsible for the funny current (If) which initiates cellular depolarization and action potential [7, 8]. The SAN is the first point of cardiac excitation and primarily responsible for allowing the heart to beat at high rates [9]. The SAN also is the main point of sympathetic nervous system regulation of the heart [10]. Electrical signals from the SAN propagate from the 2 SAN location at the superior vena cava (SVC)-right atrium (RA) junction within the terminal groove of the RA through atrial tissue or along specialized projections of the SAN called internodal tracks. These signals propagate to the AVN [11-13]. The AVN is a complex structure of over five pacemaker cell types that serves as a relay station for electrical signals to the ventricles of the heart [14]. It is located within the triangle of Koch region at the septal base of the RA [13, 15]. The AVN is responsible for slowing down electrical signals before they continue to the HB and bundle branches. This is accomplished by passing electrical signals through a network of cells that boasts a wide variety of connexin, HCN, and other ion channel concentrations [11, 14, 16-18]. Slowing down conduction just before the fast conduction pathway of the HB and bundle branches is a crucial step to achieving uniform ventricular contraction. Structure and microstructure characterization of the SAN and AVN began in the early 1900s with Keith and Flack’s work on the SAN, and Tawara’s study of the AVN [19, 20]. Keith and Flack describe the then termed ‘sino-auricle node’ as a site of differentiation in atrial muscle and connective tissue where the rhythm of the heart begins. They provided sketches of heart structures and sections of the SAN. Since that study, multiple advances were made that allowed researchers to stain and image serial sections of the SAN. Histological images of the SAN, like Masson’s trichrome images, revealed the exact location, association with adjacent structures, and microstructure of the SAN as early as the 1970s [21]. Since that time, combining histological staining with more advanced immunohistochemistry, scanning electron microscopy, and computer reconstruction led to vast understanding of the structure and microstructure of the SAN [22-25]. It is now well established that the human and large mammal SAN is approximately 3 2 cm long, 0.5 cm wide, and 0.5 cm deep. This SAN follows the epicardial terminal groove at the SVC-RA junction. It is coupled to a paranodal region that aids in the SAN’s ability to maintain a fast-acting source-sink relationship and as a result pace the heart at fast rates [26]. In Masson's trichrome images, the SAN appears as a compact blue structure within the terminal groove, superficial to atrial muscle. High-resolution imaging has shown the structure to be comprised of small, spindle-like pacemaker cells combined into bundles. These regions often surround the sinus artery and are highly innervated [22, 24]. In 1906, Tawara conducted anatomic and histological studies that defined the atrioventricular conduction pathway in the mammalian heart and identified the AVN, HB, LBB and RBB, and the PF. His work on the AVN identified multiple cell types, predicted slow conduction through the compact node, and identified AVN extensions that feed into the compact AVN body. Tawara’s findings were the basis for interpreting electrocardiography and many other advances in the field of cardiology [20, 27]. Since his work, advanced histological staining, cell isolation, immunohistochemistry, and computer reconstruction continued to help researchers better understand this complex structure [1416, 28-33]. The AVN originates on the coronary sinus (CS) of the triangle of Koch region as two extensions that feed into the compact AVN. These extensions carry the slow and fast pathways [14]. The compact node then penetrates into the IVS and becomes the HB. The compact nodal region also includes an inferior nodal bundle that parallels the endocardial surface in the caudal direction [20]. The AVN is surrounded by transitional cells in ringlike structures that are a transition in cell type from nodal cells to atrial muscle cells [12, 15]. In Masson’s trichrome images the AVN-HB complex has four distinct regions, 4 namely the AVN extension region, the compact AVN, the transition to HB region, and the HB. Each of these regions has a distinct microstructure, but in general, these structures appear partially insulated from atrial muscle tissue. Connective tissue volume fraction within the structures are higher than surrounding muscle, and the AVN/HB cells will be a lighter shade of trichrome red [15, 29]. Intraoperative Localization of the Cardiac Conduction System Every year hundreds of thousands of surgeries and procedures in the US take place in anatomical regions near the SAN and AVN. These procedures include repair of critical congenital heart defects, right atrium and AVN ablations, HB pacing, as well as other cardiac vasculature and valve replacement and repair operations. The success and effectiveness of these procedures is directly linked to reliable localization of the SAN and AVN. Inadvertent trauma to a node or regions nearby can lead to nodal dysfunction, which is associated with high levels of morbidity and mortality [34-37]. Depending on the type of procedure, up to 45% of cases result in trauma to the node, or CCS structure localization failure [38-43]. Since the 1970s, researchers and surgeons have sought to design methods and instrumentation that aid in localizing these structures. Unfortunately, many of these methods have not shown the sensitivity and specificity needed to improve upon general anatomical knowledge, or they have not integrated well into the operation and operating space. Surgeons are limited in their understanding and familiarity with anatomical markers for the nodes. Uncertainty near the nodal regions often leads to area avoidance and residual lesions or inadvertent trauma. Also, the specific size, shape, and location of the SAN and 5 AVN vary, especially in congenitally deformed hearts [13, 40]. For these reasons, realtime, comprehensive localization of the SAN and AVN is a crucial key to limiting nodal dysfunction outcomes. Fiber-Optic Confocal Microscopy In 2009, researchers at the University of Utah developed the idea to utilize fiberoptic confocal microscopy (FCM) as a method for localizing nodal tissue. This technology was used previously in gastrointestinal and pulmonary medicine, but never in the cardiac space. Their research established that FCM could produce real-time high-resolution images of cardiac tissue at a focal depth of approximately 60 µm in arrested and beating hearts. FCM imaging utilizes inert and safe fluorophores like fluorescein to mark the extracellular matrix of cardiac tissue, allowing researchers and surgeons to view the orientation of cardiac cells [44]. FCM showed great promise in localizing SAN and AVN tissue in small animal model studies, specifically rat [44, 45]. The University of Utah team identified distinct differences in normal working myocardium and nodal tissue. Nodal tissue was found to be much more reticulated and unorganized, and spindle-like cell bodies were visible when in the nodal region of both the SAN and AVN [44, 45]. With funding and support from the National Heart, Lung and Blood Institute, successful and safe integration into the operating room and procedures was reported [46]. The first-in-human clinical trials were FDA-approved and are now underway for this technology at Boston Children's Hospital. 6 New Ovine and Human Studies When this project moved into larger mammals, like ovine and human, immunohistochemistry and conventional confocal microscopy methods previously used to validate the sensitivity and specificity of this technology yielded different results than those seen in rat and rabbit. Immunohistochemistry proved difficult in ovine tissues, and imaging seemed to identify superficial cell networks rather than complete nodal structures. In search of a validation approach that was effective in ovine tissue, we found surgeons and pathologists preferred images of histologically stained tissue sections. As a result, we decided to validate in-ovine FCM results with transverse Masson's trichrome-stained sections of the SAN and AVN. We carried out experiments to identify tissue dissection, preparation, sectioning, and staining methodologies that allow us to analyze the spatial position of the nodes, nodal microstructure, and adjacent tissue. Upon finalizing the Masson's trichrome section protocols, we collected transverse sections from a complete SAN and AVN with a section spacing of 100 µm. These results showed potential for many computerized analyses and sparked the idea to perform the geometrical analysis of the SAN and AVN contained in Chapter 2 of this report. At its foundation, this work aimed to provide surgeons and researchers with usable information on the 3D positioning of the SAN and AVN, and the composition of the surrounding tissue. We wanted to validate FCM results from ovine and human studies and provide a validation system for new technologies and methods for intraoperative localization of the SAN and AVN. Specifically, we planned to analyze multiple ovine and human hearts to understand the depth of the nodes, how depth changes with age, what type 7 of tissue lies superficial to the node, and what was need to comprehensively image and localize these structures. This work involves the compilation of a library of neonatal and pediatric ovine and human SAN and AVN sections. Over 12 complete sets of SANs and AVNs were sectioned at every 100 µm throughout the tissue and stained via Masson's trichrome protocols. These sections were imaged using high-resolution slide scanning and analyzed using semiautomated computer programs designed to handle the often-complex geometry of the SAN and AVN. Nodal depth from the epi/endocardial surface was collected, nodal microstructure was qualitatively analyzed, types of tissue superficial and adjacent to the node were quantified. Our studies aimed at providing clinicians and researchers with a functional understanding of the region around the SAN and AVN, and define the probe depth and sensitivity need to comprehensively localize the nodes. Our work constitutes an important step to eliminating procedural inadvertent trauma to the SAN and AVN, and ensuring cardiothoracic procedures are successful and effective for those who need them. CHAPTER 2 GEOMETRICAL ANALYSIS OF THE SINOATRIAL AND ATRIOVENTRICULAR NODAL REGION IN OVINE AND HUMAN Introduction In the mammalian heart, the cardiac conduction system (CCS) is responsible for the initiation and propagation of electrical signals [1, 2, 12, 47]. The CCS consists of the sinoatrial node (SAN), the atrioventricular node (AVN), the His bundle, the left and right penetrating bundle branches (LBB and RBB), and the Purkinje fiber network. General understanding of the location of the CCS components in relation to superficial anatomical landmarks was established by histological studies [12, 23], but research has shown that the exact location and morphology of these components cannot be determined through surgical identification or gross dissection [13, 15, 48]. Furthermore, the location of these structures has been shown to vary from heart to heart, especially in congenitally deformed hearts. Dysfunction of the CCS is associated with high rates of morbidity and mortality [34-37]. CCS dysfunction was linked to cardiac diseases, e.g., myocardial ischemia and infarct, and acquired conditions like drug toxicity and complications due to surgical and interventional procedures [6, 49]. Various surgical and interventional cardiology procedures can lead to CCS dysfunction [50-51]. The procedures include reconstructive 9 surgeries for repair of critical congenital heart defects (CCHDs), ablation procedures in the right atrium (RA), lead placement for His bundle pacing, and cardiac valve replacement. The efficacy of these cardiac procedures is dependent on the precise localization of the CCS components. Avoidance of trauma to CCS components is crucial to each procedure. While the effectiveness of these procedures has improved over the years, many metrics of performance show that procedural complications due to poor CCS localization are still prevalent [28-43]. Current techniques to localize the CCS components include the identification of superficial anatomical markers, yet the CCS component locations vary in threedimensions. In addition, these locations vary in individuals depending on many factors including age, disease and genetics [48, 52]. A more complete understanding of the location of CCS components can provide the surgeon with additional insight to avoid traumatizing these regions during interventional cardiac procedures. Intraoperative imaging and probing technologies for localization of the CCS have been a goal since the early 1960s [53, 54]. Unfortunately, many of these new approaches did not show the sensitivity and specificity necessary to positively impact the outcome of these cardiac procedures. Often these technologies and approaches do not integrate well into the workflow of the procedure and the operating room. Recent work in this field with various forms of computerized tomography (CT) and fiber-optic confocal microscopy (FCM) showed promise in localizing the CCS. However, CT localization requires changes to procedural setups and relies heavily on anatomical landmarks near the CCS. FCM allows physicians to acquire real-time images of the SAN and AVN in beating or arrested hearts. FCM is safe and easily integrated into the operating space, but it has yet to be validated in 10 clinical studies [44, 46, 55, 56, 57, 58]. One step in advancing intraoperative localization of the CCS and cardiac procedure effectiveness is a reliable geometric characterization of the cardiac anatomy within the vicinity of these major CCS components. Understanding of the nodal region, or regions possibly affected by cardiac procedures, besides the approximate location of the CCS with respect to surface markers is limited. Many studies present significant findings related to the localization and composition of these structures, but this information is often difficult to translate to intraoperative localization [3-4, 12, 23, 29, 56]. For this work, we define the nodal region of the SAN includes the epicardial surface of the RA and superior vena cava (SVC) junction or terminal groove, and paranodal regions in the proximity of the crista terminalis (CT). The vicinity of the AVN studied in this work includes the RA endocardial surface that covers AVN origination, AVN transition to His bundle, His bundle, and the start of the left and right bundle branches. Here, we quantitatively analyzed the SAN and AVN regions in mammalian hearts. We focused on neonatal and juvenile human and ovine hearts. We utilized Masson Trichrome staining, high-resolution imaging and methods of digital image analysis to serially track and characterize SAN and AVN. Our analysis provides information on the spatial location of the SAN and AVN, and their variability. We quantified nodal depth in relation to heart age, and assessed tissue composition in nodal tissue regions. Our work provides high-quality serial image stacks and a methodology for validation of CCS localization technology and methods. 11 Results We analyzed serial sections from AVN and SAN regions of 8 ovine hearts. Serial sectioning of AVN regions started at the coronary sinus and ended close to the membranous septum (Fig. 1A). Sectioning of the SAN region started at the SVC-RA junction and ended near the inferior vena cava (IVC)-RA junction (Fig. 1B). An example of a Masson's trichrome image of the AVN region is shown in Fig. 1C. In this image, the node resides superficial to the central fibrous body (CFB) and follows the contours of the IVS along with the RA overlay below the RA endocardial surface. We present an example of a Masson's trichrome image of the SAN region in Fig. 1D. The SAN is located on the SVC side of the CT on the endocardial side and terminal groove (TG) on the epicardial side. The node resides beneath a fat pad and in close proximity to the paranodal region. The number of sections per AVN and SAN ranged from 11 to 24 and 35 to 75, respectively. After cropping, we segmented nodal regions and the tissue surface (Fig. 2). The tissue surface segmentation was fit to a spline function. We calculated the distance from the tissue surface as well as the SVN (Fig. 2C-D) and AVN (Fig. 2G-H) within the tissue. An overview of sections from an example AVN is shown in Fig. 3A-D. Nodal depth and coverage analysis were aided by dividing the node region into 4 anatomical regions (Fig. 3A-D), i.e. the AVN extension region (Fig. 3A), compact node region (Fig. 3B), transition to His bundle region (Fig. 3C), and the His bundle and distal components of the CCS. The left and right AVN extensions, both highly innervated (Fig. 3A), combine to form the compact AVN. The compact node (Fig. 3B) penetrates the connective tissue connecting the central fibrous body to the interventricular septum and exhibits morphology 12 of both the compact node and the His Bundle (Fig. 3C). This transition turn of the AVN ends when the His bundle is fully formed and insulated from cephalic atrial tissue (Fig. 3D). We summarize measurements of the depth of the AVN in Fig. 3E. The depth of this AVN, from endocardial surface to nodal surface, varied widely reaching a minimum of 90 µm at section #8, and a maximum of 1379 µm at section #2 (Fig. 3E). We present statistical analyses of the depth of ovine AVNs in Fig. 4. Fig. 4A provides a summary of aligned AVN minimum depth topology. The transition of AVN to His bundle was used as the alignment origin. Fig. 4B details AVN depths across the ovine hearts using box plots indicating depth mean and quartile ranges. Nonlinear regression analysis (Eq. 1) on data from Fig. 4B revealed a logarithmic increase in nodal depth with age (R2 value of 0.8182 and a p value of 8.41E-5). Regression analysis of the complete set of AVN depths (Fig. 4C) yielded coefficients [-102.607, 135.018]. Depth analyses of the 3 AVN anatomical regions are presented in Fig. 4D-I. Fig. 4D and G show depth in the AVN extension region and nonlinear regression, respectively. The regression analysis of Fig. 4G yielded coefficients [-56.522, 133.434], had an R2 value of 0.4920, and a p value of 2.10E-3. Fig. 4E and H show the depth and nonlinear regression analysis of the compact node region. This regression analysis yielded coefficients [-47.525, 118.613], had an R2 value of 0.6170, and a p value of 5.60E-4. Fig. 4F and I detail the depth range and nonlinear regression of the transition to His bundle region. This regression analysis yielded coefficients [137.854, 83.820], had an R2 value of 0.2260, and a p value of 9.60E-2. We analyzed the surrounding and superficial tissue in AVN and SAN Masson’s trichrome sections to characterize nodal coverage. Ovine AVN sections, like the example 13 from Fig. 5A, were divided into 3 section regions, namely the nodal region or nodal projection region as well as the cephalic and caudal regions adjacent to the nodal region (Fig. 5B). Dual red thresholding and mapping of circle regions of interest allowed for the quantification of a trichrome red percentage along the length of the sections (Fig. 5B). Example regions of interest with their trichrome red content are shown in Fig. 5C-E. Comparison of trichrome red percentages and nodal depth are presented in Fig. 5F and G. The trichrome red percentage was collected for each of the three section regions (Fig. 5H). We found significant difference of red percentage between all three groups (Group 1: Group 2 p value = 9.90E-4, 1:3 p value = 9.56E-10, 2:3 p value = 1.30E-4). Linear regression of trichrome red percentage within a section region group revealed a negative relationship between the red percentage and age in each region (Fig. 5I). Linear regression analysis (y = -0.383x + 42.039, R2 = 0.510, p value = 4.70E-2) of the complete animal red percentage mean values (Fig. 5J) illustrates a total decrease in trichrome red percentage of over 10% across the age span of the hearts in this study. Ovine SAN depth analyses followed a similar process as AVN sections. SAN section minima across the eight animals are shown in Fig. 6A. Linear and nonlinear regression analysis of this complete data set yielded no significant trend in depth vs. age. In contrast to AVN trichrome red percentage quantification, we analyzed trichrome white coverage of the SAN sections to identify adipose tissue. Trichrome white percentages for regions of interest in sections from each animal make up Fig. 6B. Analysis of the relationship between SAN depth and trichrome white coverage yielded no significant pattern or results. To augment the data and images compiled from transverse sectioning, sectioning 14 parallel to the surface was also performed. Example sections from this method are shown in Fig. 7. This mode of sectioning allowed for confirmation of nodal depth and coverage analyses previously performed and show the spatial location of the CCS structures from the normal view of physicians. Fig. 7A illustrates the four regions of the AVN and His bundle complex previously shown in Fig. 3A-D at a depth of approximately 300 µm. The tissue morphology of each region of the AVN is also shown in greater detail in Fig. 7B. Visualization of the area superficial to the AVN and His bundle structure revealed reticulated atrial overlay cells that lie directly on top of the AVN extension and compact node regions (Fig. 7C-D). Surface parallel sectioning of SAN tissue is presented in Fig. 7E. The morphology of the SAN region (Fig. 7F) matches segmented nodal tissue in transverse sectioning. We also studied AVN and SAN regions in human neonatal hearts. Nodal depth analysis was performed on four neonatal human AVNs and three neonatal human SANs. Example images of transverse sections from the neonatal hearts are shown in Fig. 8A and B. The complete set of human AVN depth measurements makes up Fig. 8C. Human AVN depth ranged from 19.4 µm to 902.1 µm. Fig. 8D reveals the nonlinear regression analysis of human AVN depth minima ([-1260.700, 662.202], R2 = 0.877, p value = 2.30E-2). We found that human AVN depth, like ovine AVN depth, has a logarithmic relationship with age. Human SAN depth ranged from 4.9 µm to 716.7 µm. SAN depths is depicted in Fig. 8E. Like ovine SAN depth analysis previously described, the relationship between SAN depth vs. age was weak (R2 = 0.015, p value = 1.67E-1). We did not find significant epicardial adipose tissue proximal to the SAN. 15 Discussion In this investigation, we performed a geometrical analysis of the AVN and SAN that can help in validation of methods and instrumentation for the localization of the CCS. We collected AVN and SAN regions from 8 ovine and four human hearts and performed depth and morphology analyses on serial Masson's trichrome sections. Results from 170 AVN sections and 490 SAN sections were analyzed from hearts that ranged from a preterm 29-week-old human to a 1-month-old human, as well as a preterm 18-week-old sheep to a 49-month-old sheep. Using CCS microstructure identification guidelines well established over the past century [12, 15, 19, 20, 22, 59], we segmented high-resolution images collected every 100 µm through complete or near complete AVNs and SANs (Fig. 2). Transverse serial sectioning of ovine and human AVN tissue confirmed the anatomical existence of the left and right AVN extensions in both species [14]. AVN extension tissue appeared much more innervated than compact AVN tissue (Fig. 3A-B). Also, we observed distinct tissue microstructure change from the compact node region to the His bundle (Fig. 3C vs. 7B). AVN depth analysis results show a similar node depth topology across all age groups in this study (Fig. 4A). Surgeons and researchers can expect the membranous septum side of the AVN extensions and their fusion into compact node body to be the most superficial portion of the AVN and His Bundle complex (Fig 3E and 4D-E). The origination of the AVN extensions and the transition to His Bundle region mark sites where depth increases significantly (Figs. 3E and 4F). Our work revealed that minimum AVN depth increases logarithmically with age (Fig. 4). This increase in depth occurred in each of the 3 AVN anatomical regions (Fig. 16 4G-I) and for the complete node (Fig. 4C). While the logarithmic fit of the increase in depth at the transition to His Bundle region did not meet our definition of statistical significance, we believe in light of the global increase in depth this transition to the His bundle should be expected to follow a similar trend. Based on this finding, we suggest procedural methods and instrumentation should be accurately adapted to the age of the patient. Suture bite depth, imaging protocols, and many other procedure elements should change accordingly with age. The measured average and minimum depths of AVNs suggest that localization imaging techniques require an imaging depth of over 300 and 400 µm to comprehensively image the AVN from origination to transition to His Bundle in human and ovine, respectively. There are, however, points of minimum depth in each AVN that would require a much smaller imaging depth for localization of a single point. Our study demonstrated, through the quantification of trichrome red percentages in tissue superficial to the AVN and imaging of sections parallel to the endocardial surface, that it is possible to use superficial tissue composition and microstructure to identify deeper nodal structures. Superficial tissue, or AVN coverage, has significantly different trichrome red percentages between tissue cephalic or caudal to the AVN and the actual AVN nodal region (Fig. 5). Our dual red thresholding method for this work allowed us to relate the trichrome image red percentage to the actual cardiac myocyte volume fraction within these complex tissue sections. By detecting variations in cardiac myocyte volume fraction around the Triangle of Koch, surgeons and researchers will be able to accurately localize deeper nodal structures. The decrease in cardiac myocyte volume fraction superficial to the AVN is due to the thinning of the atrial overlay and higher collagen percentages within 17 networks of AVN transitional cells [60-62]. The accuracy of this method is further enhanced by the significant decrease in cardiac myocyte volume fraction caudal to the node (Fig. 5H). In addition to changes in superficial tissue composition, distinct cellular networks superficial to the node could serve as reliable markers for deeper AVN structures. These cellular networks are the termination of the atrial overlay (Fig. 7C-D). We discovered these superficial structures by sectioning AVN tissue parallel to the endocardial surface and taking great care to preserve, stain, and image the most superficial sections. The reticulated pattern of the atrial overlay termination lies superficial to the fusion of the AVN extensions and the compact node. These structures were specifically identified previously with fiberoptic confocal microscopy in human and ovine hearts [46], and have been further validated in this work. The depth and coverage analysis methods of this work were designed to successfully characterize the complex geometry of the SAN. The SAN has been shown to reside near the surface of the epicardium on the SVC side of the terminal groove [13, 22]. Our work confirms that the minimum depth of SAN from the epicardial surface ranges from 0.1 to 25.7 µm in ovine, and 4.5 to 40 µm in human. Our regression analysis shows that there is no relationship between the depth of the SAN and age (Figs. 6A and 8E). While portions of the SAN are often visible with the naked eye [13], comprehensive localization of the SAN will require an imaging depth of approximately 115 µm (Fig 8E). In many SAN sections we imaged, the SAN is not limited to a compact structure at the terminal groove. Rather, there are many projections of the node (Fig. 8B) that can reach up to 5mm on both sides of the terminal groove. Imaging of these structures should yield images of 18 microstructure similar to that seen in Masson’s trichrome images in this report (Figs. 1D, 2H, 7F and 8B). In an attempt to characterize the relationship of SAN coverage, or the covering of the SAN by adipose tissue, and SAN depth, we quantified the trichome white percentage in tissue superficial to the node. This analysis showed that trichrome white percentage varies widely from section to section and there is no relationship between the coverage and depth. One cause of this wide range of Masson's trichrome white percentage and the poor relation to depth include the presence of nodal projections that stretch to the boundaries of fat pads. The results of human AVN and SAN depth analyses were similar to the results we found in ovine. Specifically, the depth of the AVN in human also increased in a logarithmic manner with age (Fig. 8D). This finding, coupled with the superficial nature and similar depth of human SAN, leads us to suggest that AVN depth results fully characterized by our ovine portion of this study accurately apply to procedures performed on humans. Again, we advise that procedures and studies performed in the vicinity of the AVN should alter methodologies and instrumentation according to the age of the patient. While the SAN did not show age or species dependent depth changes, we found that SAN localization techniques are crucial to avoid trauma to the SAN as the location and morphology of the node changed from subject to subject. Limitations This work focused on neonatal and juvenile hearts from human and ovine. As a result, this work lacks data from ovine hearts between the ages of 7.1 months and 53.4 19 months, and human hearts older than one month of age. To extend the age range and scope of this study, we included one adult ovine heart (53.4 months old). Analysis of the relationship of depth vs. age including the older ovine further supported a nonlinear regression model. Without the inclusion of this older heart in the analysis, the neonatal and pediatric ovine hearts still followed a statistically significant nonlinear regression of a logarithmic model ([-132.04, 150.16], R2 = 0.5654, p value = 1.60E-3]. Furthermore, ovine AVN coverage between the three section regions was still significantly different (1:2 p value = 9.87E-4, 1:3 p value = 9.56E-10, 2:3 p value = 1.27E-4). Analyses in our study were performed on images from sequential 2D sections. We limited our analyses to sections without artifacts due to shearing, tearing, or nonuniform immersion in staining solutions. Processing and analyses were limited to 2D images from section collected a high spatial resolution, i.e., a spacing of 100 µm. Alternatively, 3D analyses would be feasible through computational reconstruction of the AVN and SAN tissue. Methods Heart Collection and Sources Ovine hearts were obtained through IACUC approved protocols at Boston Children's Hospital and the University of Utah. Animal ages ranged from 12 hours to 4 years four months. After euthanasia of the animals, hearts were excised and perfused with a cold, zero-calcium Tyrode solution (in mmol/L: 92 NaCl, 11 dextrose, 4.4 KCl, 5 MgCl2, 24 HEPES, 20 taurine, 5 creatine, 5 C3H3NaO3, 1 NaH2PO4, 12.5 NaOH; pH 7.2; ≈10°C). Subsequently, the heart was fixed using the Tyrode solution containing 4% 20 paraformaldehyde via antegrade perfusion with a syringe. The hearts were then immersed in the same solution for up to 24 h before being stored in phosphate buffered saline (PBS) until dissection. Human heart studies were granted Institutional Review Board exemption at the University of Utah (IRB_00111846) and Boston Children's Hospital. We obtained two deidentified, post-autopsy hearts from preterm live birth mortality cases. Gestational ages were 29 weeks 1 day and 38 weeks 6 days. SAN and AVN regions were preserved during an autopsy. We also obtained de-identified, formalin-fixed sections from the SAN, AVN, and His Bundle regions of 2 hearts without congenital defects. Ages ranged from 12 days to 1 month. Node Dissection SAN and AVN regions were dissected from each heart using anatomical landmarks [8]. Excised SAN tissue samples included the crista terminalis of the RA and stretched from 2-4 mm beyond the crest of the RA appendage in the cephalic direction to the RAInferior Vena Cava (IVC) junction. At least 5 mm of tissue was included on both the SVC and RA appendage (RAA) sides of the crista terminalis. Excised SAN samples were approximately 1 cm wide (SVC to RA appendage across the terminal groove), 3 cm long (SVC-RA junction to IVC-RA junction) and included both the epicardial and endocardial surface of the RA. AVN excised tissue samples spanned from proximal to the coronary sinus (CS) ostium to the membranous septum (MS), and cephalic to the tendon of Todaro to below the annulus of the tricuspid valve septal leaflet. We ensured that the entire Triangle of Koch and cephalic portions of the RBB and LBB were included in the samples. 21 In human hearts, AVN samples were approximately 1 cm by 1 cm and included the left atrial (LA) and left ventricular (LV) side of the septum. In ovine hearts, AVN samples ranged from 1 cm by 1 cm to 3 cm by 3 cm and included the LA/LV side of the septum. Histology Section Preparation and Imaging Excised tissue samples were marked for identification of the sectioning face with a tissue dye that withstood paraffinization (Cancer Diagnostics, Inc.). Samples were then dehydrated through a gradient of ethanol and deionized water solutions which ended in 100% ethanol. After dehydration, the tissue was cleared with a gradient of a paraffinmiscible organic solvent called Citrisolv and ethanol solutions. Clearing ended with immersion in 100% Citrisolv. The tissue was then immersed in paraffin solutions with decreasing amounts of Citrisolv and placed in a heated vacuum container at 60 °C. Tissue was completely infiltrated by paraffin wax after two immersion steps in 100% paraffin. Times for each step varied based on tissue sample size. Paraffinized tissue was then placed in a paraffin mold and oriented so that the marked sectioning face of the tissue was accessible. Paraffin blocks hardened overnight. Sectioning was performed on a HistoCore BIOCUT microtome (Leica Biosystems, Wetzlar, Germany). Paraffin blocks were softened in a solution of ice water and 16.67% fabric softener by volume to minimize section tearing. We collected 8-10 serial sections with a thickness of 4 µm at every 100 µm throughout each tissue sample. AVN samples were sectioned from CS to MS. SAN samples were sectioned from RA-SVC junction to RA-IVC junction along the crista terminalis. Sections were collected on positively charged tissue adhesion slides and placed on a slide warmer until paraffinized sections were flush 22 across the slide. Completed slides were stored at room temperature. SAN and AVN samples from one ovine heart were also sectioned parallel to the epicardial and endocardial tissue surface, respectively. Half of the collected sections were stained using an automated Masson’s Trichrome protocol on an automated staining device (Dako Corporation, Carpinteria, CA). Stain immersion duration and differentiation were verified using liver tissue sections of the same thickness every 16 sections. Imaging of the serial sections was performed on an AXIOSCAN Z.1 slide scanner (Cark Zeiss AG, Oberkochen, Germany) using a 40x objective as well as automated section identification and focus protocols. Image post-processing, including color balance adjustment and rotation, were performed using Fiji [62]. Image Segmentation and Geometrical Analyses Manual segmentation using MATLAB (Mathworks, Natick, MA) identified nodal tissues and the epi/endocardial surface in each section. The segmentations were used to analyze the morphology, depth, and coverage of nodal tissue in each section. The segmented epi/endocardial surface was fit to a smoothing spline function using the MATLAB fit function. Distance maps were generated in MATLAB to calculate the distance from both the node and surface. Nodal depth was determined as the values of the nodal distance map along the fitted surface function points. Full-length nodal depth vectors and a minimum nodal depth were recorded for each section. Node associated sub-epi/endocardium tissue that is superficial to the node (nodal coverage) was analyzed in both SAN and AVN. Due to the superficial localization of the 23 SAN, epicardial adipose tissue was measured using a white content threshold on the Masson’s Trichrome images. Coverage was recorded in conjunction with nodal depth as a white content percentage of superficial tissue to any nodal tissue. AVN nodal coverage, or sub-endocardium working myocardium, was measured using a red content dual threshold on the Masson’s Trichrome images. Again, coverage of AVN tissue was recorded with nodal depth data as a red content percentage of tissue superficial to any nodal tissue. Coverage percentages were determined by centering a series of circle regions of interest on the fitted surface function. SAN circles had a radius of 150µm and were spaced along the surface every 75µm. AVN circles had a radius of 500µm and were spaced along the surface every 250µm. The trichrome image white or red within these regions of interest that resided deep to the surface mask and superficial to the nodal mask was averaged and reported as a percentage. Statistical Analyses Nodal depth and coverage were reported as mean +/- standard deviation. Standard box and whisker plots, depicting the mean, quartile ranges, and outliers, were used to illustrate data from the ovine and human hearts. Linear and nonlinear regression models were created with the MATLAB functions fitlm and fitnlm. The nonlinear model was defined as: y(age) = a + b log(age). (1) The models were considered significant compared with respective models if F statistics yielded a value of p < 0.05. The coefficient of determination R2 and model p values determined the goodness of fit for each of the regression models. ANOVA with post 24 hoc Tukey-Kramer test was performed on measures of AVN sections to determine significant differences. 25 Figure 1: Sectioning and Masson’s trichrome imaging of SAN and AVN. (A) Endocardial view on ovine AVN and (B) epicardial view on SAN region with sectioning orientation and direction indicated by red lines and arrow, respectively. Scale bars: 1 cm. Example Masson’s trichrome stained ovine (C) AVN and (D) transverse section. Scale bars: 1 mm. Nodes are outlined in green. SAN paranodal region outlined in orange. Sutures in panels A and B represent approximate node locations identified during a localization study. CS, coronary sinus; MS, membranous septum; TV, septal leaflet of the tricuspid valve; RAA, right atrial appendage; TG, terminal groove; SVC, superior vena cava; IVC, inferior vena cava; RA, right atrium; CFB, central fibrous body; IVS, interventricular septum; CT, crista terminalis. 26 Figure 2: Analyses of Masson’s trichrome images. (A-D) Segmentation process for AVN. (A) Cropped transverse section of ovine AVN region. Scale bar: 500 µm. (B) Segmentation of endocardial surface and AVN, indicated in white. (C) Contour plot of distance map from the endocardial surface with line spacing of 500 µm overlaid on (A). D, Contour plot of distance map from node surface with line spacing of 500 µm overlaid on (A). (E-H) Segmentation process for SAN. (E) Cropped transverse section of ovine SAN region. Scale bar: 500 µm. (F) Segmentation of epicardial surface and SAN, indicated in white. (C) Contour plot of distance map from the epicardial surface with line spacing of 500 µm overlaid on (E). (G) Contour plot of distance map from node surface with line spacing of 500 µm overlaid on (E). RA, right atrium; CFB, central fibrous body; CT, crista terminalis. 27 Figure 3: Masson’s trichrome images and depth analyses from 4 different anatomical regions of the AVN-His bundle complex. (A) Cropped transverse section of ovine AVN extension region with separation of left and right AVN extensions. (B) Cropped transverse section of ovine AVN compact node region. (C) Cropped transverse section of ovine AVN transition to His bundle region. (D) Cropped transverse section of ovine His bundle. (E) Depth profile of AVN from RA endocardial surface. Box plots represent the range of depths for each section along the projection of the node onto the endocardial surface. Sections were collected every 200 µm. Example images from each anatomical region are shown in (A-D). Scale bars: 500 µm. CFB, central fibrous body; IVS, interventricular septum; TV, septal leaflet of the tricuspid valve. 28 Figure 4: Depth analyses of ovine AVN. (A) Aligned AVN depth profiles from the RA endocardial surface for eight ovine hearts. Postmenstrual age of each animal is shown in the key. (B) AVN section minimum depths. (C) Nonlinear regression model of relationship of AVN depth and age (R2=0.8176, p value = 8.12E-5). Section minimum depth with log fit. (D) AVN extension region section minimum depths. Only sections that occurred before the formation of the compact node were measured for this data set. (E) AVN compact node region section minimum depths for each heart. The analysis was limited to sections that before the transition to His bundle. (F) AVN transition to His bundle region section minimum depths. Only sections before the His bundle were measured for this data set. (G) Nonlinear regression model for relationship between AVN extension region depth and age (R2=0.4920, p value = 2.31E-3). Region section minimum depth data set shown with the log fit of animal minima. (H) Nonlinear regression model for relationship between AVN compact node region depth and age (R2=0.6168, p value = 5.60E-4). Region section minimum depth data set shown with the log fit of animal minima. (I) Nonlinear regression model for an increase in AVN transition to His bundle region depth with age (R2=0.2255, p value = 5.60E-4). Region section minimum depth data set shown with the log fit of animal minima. 29 Figure 5: Coverage analyses of ovine AVN. (A) Masson's trichrome stained transverse section with outlined compact node. Scale bar: 1 mm. (B) Red threshold image of (A) with representative coverage quantification circles plotted at half the original density. Yellow stars indicate division points along the endocardial surface between the nodal region as well as the cephalic (top) and caudal (bottom) regions of the section. Green arrows indicate regions of interest displayed in (CE). (C-E) Red threshold image in the region of interest corresponding to the top, middle and bottom green arrow, respectively in (B). Gray areas indicate regions not part of the analysis. (F) AVN coverage of Masson's trichrome red percentage from each of the circle regions of interest mapped to the section surface. Green arrows indicate the percentage data point that corresponds to the analysis of the circle regions of interest depicted in (C-E). (G) Depth of node beneath the endocardial surface. Red sections of F and G plots represent the nodal section region. (H) Masson’s trichrome red percentage for ovine AVNs divided into the three section regions. Red percentages are different in-between groups (1:2 p value = 9.90E-4, 1:3 p value = 9.56E-10, 2:3 p value = 1.30E-4). (I) Linear fit slope for each section region's Masson's trichrome red percentage data with respect to age, showing a decrease in Masson's trichrome red percentage with an increase in animal age. (J) Masson's trichrome red percentage with linear regression fit to animal mean (R2 = 0.5102; p value = 4.65E-2). 30 Figure 6: Depth and coverage analyses of ovine SAN. (A) Minimum depths of sections. (B) Trichrome white percentage from SAN section regions of interest. 31 Figure 7: Example Masson’s trichrome sections of ovine AVN and SAN. (A) AVN section parallel to the endocardial surface at a depth of approximately 300 µm. Scale bar: 5 mm. (B) Zoomed region from green box in (A) with AVN left and right extensions, compact node, and the His bundle. Scale bar: 500 µm. (C) AVN section parallel to the endocardial surface at a depth of 50 µm showing the terminal region of the atrial overlay. Green box and scale match (A). Scale bar: 5 mm. (D) Zoomed region from yellow box in (C) with reticulated overlay cells. (E) SAN section parallel to the epicardial surface at a depth of approximately 100 µm with terminal groove and SAN. Yellow lines mark the boundary of the terminal groove. Scale bar: 2mm. (F) Zoomed region of from green box in (E). Scale bar: 500 µm. CS, coronary sinus; RA, right atrium; MS, membranous septum; CFB, central fibrous body; IVS, interventricular septum; TV, septal leaflet of the tricuspid valve; LE, left nodal extension; RE, right nodal extension; CN, compact node; His, His Bundle; SVC, superior vena cava; RAA, right atrial appendage. 32 Figure 8: Example Masson’s trichrome sections and depth analyses of AVN and SAN in human neonatal heart. (A) Section of AVN region with compact AVN outlined in green. Scale bar: 1 mm. (B) Section of SAN. SAN outlined in green. Paranodal region outlined in orange. Scale bar: 1 mm. (C) Minimum depths of AVN sections. (D) Nonlinear regression model of relationship of AVN depth and age (R2=0.877, p value = 2.30E-2). (E) SAN section minimum depths. MV, anterior leaflet of the mitral valve; CFB, central fibrous body; RA, right atrium; IVS, interventricular septum; TV, septal leaflet of the tricuspid valve; RAA, right atrial appendage; SVC, superior vena cava; CT, crista terminalis. CHAPTER 3 CONCLUSIONS We analyzed the depth and positioning of the SAN and AVN in ovine and human hearts, as well as characterized the composition and microstructure of neighboring tissues. We sectioned and analyzed eight ovine SANs and AVNs, three human SANs, and 4 human AVNs. We achieved section and image quality that exceeds that of the field standard and those published in the past. The main findings from these analyses help to define the necessary probe depth and method target to comprehensively image or localize the SAN and AVN within an operation or procedure. We found that the ovine AVN is located within the triangle of Koch region of the RA at an average minimum depth of 171.4 µm. Human AVN positioning in neonatal hearts has a similar spatial location but resides at an average minimum depth of 109.2 µm. The AVN of both species becomes deeper in the RA endocardium with age, following a logarithmic regression model. This increase in depth was seen for all anatomical regions of the AVN-HB complex including the AVN extensions, compact nodal body, the transition to HB, and the HB. We concluded that comprehensive imaging or localization from the RA endocardial surface will require a probe depth of over 300 and 400 µm for human and ovine respectively. Ovine SAN was found to reside near the terminal groove on the RA epicardial 34 surface at an average minimum depth of 96.4 µm. Many portions of the node were much more superficial than this average minimum depth, and the SAN often appears to reside right on the epicardial surface. The human SAN in neonates had an average minimum depth of 14.8 µm. Neither species showed nodal depth changes with age. We did find however that the location and nodal projections reaching into the RA significantly varied from heart to heart. This finding indicated a need for comprehensive localization techniques. Our study also showed significant differences in the cardiac myocyte volume fraction between tissue directly superficial to the AVN and tissue cephalic or caudal to the nodal region. This result suggests that by simply detecting changes in the composition of superficial tissue, researchers and surgeons can reliably detect deeper nodal structures. In addition to these composition changes, our study revealed that there are specific microstructures superficial to the AVN that accurately mark the AVN extension and compact node regions. These superficial markers are the terminating ends of the atrial overlay, an extension of the atrial working myocardium which descends toward the septal leaflet of the tricuspid valve and covers the AVN. The cells at the termination of the atrial overlay are very reticulated and appear spindle-like in shape. While the average minimum depth of ovine and human AVNs is greater than the focal distance of FCM (60 µm), we believe that FCM can image superficially reticulated atrial overlay cells that mark the deeper AVN structure. Indeed, in many analyzed nodes FCM detected nodal tissue, but it is more likely that the AVN region was localized through the detection of reticulated atrial overlay cells. Future work on this project will thoroughly explore the reliability of the atrial overlay cells as a marker for the AVN. 35 SAN depth analysis coupled with an analysis of superficial tissue suggests that FCM can image a majority of the SAN structure in ovine and human hearts. However, regions of the SAN were covered by thick layers of adipose tissue or the caudal end of the SAN that tapers deeper into the epicardium. These regions are outside the focal depth of FCM. In order to comprehensively image or localize the SAN ovine and human, devices would need a probe depth greater than 115 µm. Our study focused on 2D sections and images taken at a high spatial resolution through complete SANs and AVNs. This methodology introduces some artifacts in the tissue and images due to tissue shearing and tearing, stain nonuniformity, and actual node depth measurements. The next step in this work will be to computationally register and reconstruct these nodes from the hundreds of 2D images that we collected. This 3D analysis will not only validate this work, but provide us with new information on node topology. Another future aspect of this project will be to include pathological tissue in the analyses. Collecting hearts with congenital defects or nodal dysfunction will be a crucial step to defining how comprehensive intraoperative localization of the nodes will be accomplished. Including other CCS structures like the HB and bundle branches will also be important if we are to take the guesswork out of these procedures. We believe the results from this work are a substantial step in not only validating FCM, but also providing the groundwork for future technologies that will image or localize CCS structures more effectively and efficiently. Our work also adds a dimension on the location of the SAN and AVN. We suggest that depth and coverage data contained in this 36 thesis will aid in adjusting cardiac procedure approaches based on patient age, and the 3D SAN and AVN structure. REFERENCES 1. Boyett, M.R., "And the beat goes on." The cardiac conduction system: the wiring system of the heart. Exp Physiol, 2009, 94(10): p. 1035-49. 2. Park, D.S., Fishman, G.I., The cardiac conduction system. Circ, 2011, 123(8): p. 904-15. 3. Park, D.S., Fishman, G.I., Development and function of the cardiac conduction system in health and disease. J Cardiovasc Dev Dis, 2017, 4(2). 4. van Weerd, J.H., Christoffels, V.M., The formation and function of the cardiac conduction system. Dev, 2016, 143(2): p. 197-210. 5. Masson-Pevet, M.A., et al., Pacemaker cell types in the rabbit sinus node: a correlative ultrastructural and electrophysiological study. J Mol Cell Cardiol, 1984, 16(1): p. 53-63. 6. Choudhury, M., Morris, G., Biology of the sinus node and its disease. Arrhythm Electrophysiol Rev, 2015, 4(1): p. 28-34. 7. DiFrancesco, D., The role of the funny current in pacemaker activity. Circ Res, 2010, 106(3): p. 434-46. 8. DiFrancesco, D., Borer, J.S., The funny current: cellular basis for the control of heart rate. Drugs, 2007, 67 Suppl 2: p. 15-24. 9. Fabbri, A., et al., Computational analysis of the human sinus node action potential: model development and effects of mutations. J Physiol, 2017, 595(7): p. 2365-2396. 10. Gordan, R., Gwathmey, J.K., Xie, L.H., Autonomic and endocrine control of cardiovascular function. World J Cardiol, 2015, 7(4): p. 204-14. 11. Atkinson, A.J., et al., Functional, anatomical, and molecular investigation of the cardiac conduction system and arrhythmogenic atrioventricular ring tissue in the rat heart. J Am Heart Assoc, 2013, 2(6): p. e000246. 12. Dobrzynski, H., et al., Structure, function and clinical relevance of the cardiac conduction system, including the atrioventricular ring and outflow tract tissues. 38 Pharmacol Ther, 2013, 139(2): p. 260-88. 13. Anderson, R.H., Hlavacek, A.M., Cook, A.C., Backer, C.L., Wilcox’s surgical anatomy of the heart. Fourth ed. 2013, New York, NY: Cambridge University Press. 14. Kurian, T., et al., Anatomy and electrophysiology of the human AV node. Pacing Clin Electrophysiol, 2010, 33(6): p. 754-62. 15. Anderson, R.H., et al., The anatomy of the cardiac conduction system. Clin Anat, 2009, 22(1): p. 99-113. 16. Greener, I.D., et al., Molecular architecture of the human specialised atrioventricular conduction axis. J Mol Cell Cardiol, 2011, 50(4): p. 642-51. 17. Hucker, W.J., et al., Connexin 43 expression delineates two discrete pathways in the human atrioventricular junction. Anat Rec (Hoboken), 2008, 291(2): p. 20415. 18. Yoo, S., et al., Localization of Na+ channel isoforms at the atrioventricular junction and atrioventricular node in the rat. Circulation, 2006, 114(13): p. 136071. 19. Keith, A., The form and nature of the muscular connections between the primary divisions of the vertebrate heart. J Anat Physiol, 1907, 41(3): p. 172-189. 20. Tawara, S., Das Reizleitungssytem des Saeugetierherzens. 1906, Author. 21. Evans, R., Shaw, D.B., Pathological studies in sinoatrial disorder (sick sinus syndrome). Br Heart J, 1977, 39(7): p. 778-86. 22. Chandler, N., et al., Computer three-dimensional anatomical reconstruction of the human sinus node and a novel paranodal area. Anat Rec (Hoboken), 2011, 294(6): p. 970-9. 23. Chandler, N.J., et al., Molecular architecture of the human sinus node: insights into the function of the cardiac pacemaker. Circulation, 2009, 119(12): p. 156275. 24. Csepe, T.A., et al., Human sinoatrial node structure: 3D microanatomy of sinoatrial conduction pathways. Prog Biophys Mol Biol, 2016, 120(1-3): p. 16478. 25. Mandrioli, D., et al., SEM, TEM, and IHC analysis of the sinus node and its implications for the cardiac conduction system. Anat Res Int, 2013, 2013: p. 39 961459. 26. Unudurthi, S.D., Wolf, R.M., Hund, T.J., Role of sinoatrial node architecture in maintaining a balanced source-sink relationship and synchronous cardiac pacemaking. Front Physiol, 2014, 5: p. 446. 27. Suma, K., Sunao Tawara: a father of modern cardiology. Pacing Clin Electrophysiol, 2001, 24(1): p. 88-96. 28. Dobrzynski, H., et al., Site of origin and molecular substrate of atrioventricular junctional rhythm in the rabbit heart. Circ Res, 2003, 93(11): p. 1102-10. 29. Duan, D., et al., Morphological study of the atrioventricular conduction system and Purkinje fibers in yak. J Morphol, 2017, 278(7): p. 975-986. 30. Efimov, I.R., et al., Structure-function relationship in the AV junction. FASEB J, 2004, 280(2): p. 952-65. 31. Ko, Y.S., et al., Three-dimensional reconstruction of the rabbit atrioventricular conduction axis by combining histological, desmin, and connexin mapping data. Circulation, 2004, 109(9): p. 1172-9. 32. Meijler, F.L., Morphology and electrophysiology of the mammalian atrioventricular node. Physiol Rev, 1988, 68(2): p. 608-647. 33. Ye Sheng, X., et al., Isolation and characterization of atrioventricular nodal cells from neonate rabbit heart. Circ Arrhythm Electrophysiol, 2011, 4(6): p. 936-46. 34. Elizari, M.V., Acunzo, R.S., Ferreiro, M., Hemiblocks revisited. Circ, 2007, 115(9): p. 1154-63. 35. Flaker, G., et al., Death in patients with permanent pacemakers for sick sinus syndrome. Ame Heart J, 2003, 146(5): p. 887-893. 36. Sinker, D., Goldberg, H., Sinus and A-V nodal dysfunction following myocardial infarction. J Electrocardiol, 1975, 8(3): p. 281-283. 37. Zanini R., Gallo, G., Cazzamalli, L., Bonandi, L., Dei Cas, L., Morbidity and mortality of patients with sinus node disease: comparative effects of atrial and ventricular pacing. Pacing Clin Electrophysiol, 1990, 13(12): p. 2076-2079. 38. Bogers, A.J., et al., Long term follow up after surgery in congenitally corrected transposition of the great arteries with a right ventricle in the systemic circulation. J Cardiothorac Surg, 2010, 5: p. 74. 40 39. Graham, T.P., et al., Long-term outcome in congenitally corrected transposition of the great arteries. J Am Coll Cardiol, 2000, 36(1): p. 255-261. 40. Gross, G.J., Hamilton, R.M., Kirsh, J.A., Stephenson, E.A., Natrural history of postoperative heart block in congenital heart disease: implications for pacing intervention. Heart Rhythm, 2006, 3(5): p. 601-604. 41. Merin, O., Oren, A., Fink, D., Deeb, M., Bitran, D., Silberman, S., Permanent pacemaker implantation following cardiac surgery: indications and long-term follow-up. Pacing Clin Electrophysiol, 2009, 32(1): p. 7-12. 42. Ruaengsri, C., et al., The Cox-maze IV procedure in its second decade: still the gold standard? Eur J Cardiothorac Surg, 2018, 53(suppl_1): p. i19-i25. 43. Vijayaraman, P., et al., Imaging evaluation of implantation site of permanent direct His bundle pacing lead. Heart Rhythm, 2014, 11(3): p. 529-30. 44. Huang, C., et al., Identification of nodal tissue in the living heart using rapid scanning fiber-optics confocal microscopy and extracellular fluorophores. Circ Cardiovasc Imaging, 2013, 6(5): p. 739-46. 45. Huang, C., et al., Sensitivity and specificity of cardiac tissue discrimination using fiber-optics confocal microscopy. PLoS One, 2016, 11(1): p. e0147667. 46. Mondal, A., et al., An imaging protocol to discriminate specialized conduction tissue during congenital heart surgery. Semin Thorac Cardiovasc Surg, 2019, 31(3): p. 537-546. 47. Nikolaidou, T., et al., Structure-function relationship in the sinus and atrioventricular nodes. Pediatr Cardiol, 2012, 33(6): p. 890-9. 48. Aiello, V.D., Understanding the morphology of the specialized conduction tissues in congenitally malformed hearts. World J Pediatr Congenit Heart Surg, 2015, 6(2): p. 239-49. 49. Kashou, A.H., Rhythm, atrioventricular block, 2018, Treasure Island, FL: StatPearls Publishing. 50. Titus, J.L., Kirlin, J.W., Edwards, J.E., Lesions of the atrioventricular conduction system after repair of ventricular septal defect. Circ, 1963, 28: p. 82-88. 51. Pasic, M., et al., Transient sinus node dysfunction after the Cox-maze III procedure in patients with organic heart disease and chronic fixed atrial fibrillation. J Am Coll Cardiol, 1998, 32(4): p. 1040-1047. 41 52. Smith, A., et al., The diverse cardiac morphology seen in hearts with isomerism of the atrial appendages with reference to the disposition of the specialised conduction system. Cardiol Young, 2006, 16(5): p. 437-54. 53. Bernhard, W.F., A method for localization of the cardiac conduction system during open-heart surgery. N Engl J Med, 1961, 265(22): p. 1079-1083. 54. Lincoln, C., Logan-Sinclair R., Anderson R.H., A cardiac conduction monitor and probe for intraoperative identification of conduction tissue. Brit Heart J, 1979, 42: p. 339-344. 55. Shinohara, G., et al., Three Dimensional Visualization of Human Cardiac Conduction Tissue in Whole Heart Specimens by High-Resolution Phase-Contrast CT Imaging Using Synchrotron Radiation. World J Pediatr Congenit Heart Surg, 2016. 7(6): p. 700-705. 56. Stephenson, R.S., et al., High-resolution 3-Dimensional imaging of the human cardiac conduction system from microanatomy to mathematical modeling. Sci Rep, 2017. 7(1): p. 7188. 57. Venius, J., Zurauskas, E., Rotomskis, R., High-resolution imaging of the human cardiac conduction system using reflectance confocal microscopy. Tohoku J. Exper Med, 2013, 229(1): p. 67-73. 58. Xu, Y., Locating the human cardiac conduction system using a 3D model of its nutritious arteries. Sci Rep, 2017, 7(1): p. 344. 59. Kawashima, T., Sasaki, H., Gross anatomy of the human cardiac conduction system with comparative morphological and developmental implications for human application. Ann Anat, 2011, 193(1): p. 1-12. 60. Defelice, L.J., Challice, C.E., Anatomical and ultrastructural study of the electrophysiological atrioventricular node of the rabbit. Circ Res, 1969, 24: p. 457-474. 61. Patterson, E., Scherlag, B.J., Delineation of AV conduction pathways by selective surgical transection: effects on antegrade and retrograde transmission. J Interv Card Electrophysiol, 2005, 13(2): p. 95-105. 62. Schindelin, J., Fiji: an open-source platform for biological-image analysis. Nat Meth, 2012, 9(7): p. 676-82.