UUnniivveerrssiittyy ooff PPeennnnssyyllvvaanniiaa SScchhoollaarrllyyCCoommmmoonnss Technical Reports (CIS) Department of Computer & Information Science January 1995 TTrraauuMMAAPP -- IInntteeggrraattiinngg AAnnaattoommiiccaall aanndd PPhhyyssiioollooggiiccaall SSiimmuullaattiioonn ((DDiisssseerrttaattiioonn PPrrooppoossaall)) Jonathan Kaye University of Pennsylvania Follow this and additional works at: https://repository.upenn.edu/cis_reports RReeccoommmmeennddeedd CCiittaattiioonn Jonathan Kaye, "TrauMAP - Integrating Anatomical and Physiological Simulation (Dissertation Proposal)", . January 1995. University of Pennsylvania Department of Computer and Information Science Technical Report No. MS-CIS-95-29. This paper is posted at ScholarlyCommons. https://repository.upenn.edu/cis_reports/197 For more information, please contact [email protected]. TTrraauuMMAAPP -- IInntteeggrraattiinngg AAnnaattoommiiccaall aanndd PPhhyyssiioollooggiiccaall SSiimmuullaattiioonn ((DDiisssseerrttaattiioonn PPrrooppoossaall)) AAbbssttrraacctt In trauma, many injuries impact anatomical structures, which may in turn affect physiological processes - not only those processes within the structure, but also ones occurring in physical proximity to them. Our goal with this research is to model mechanical interactions of different body systems and their impingement on underlying physiological processes. We are particularly concerned with pathological situations in which body system functions that normally do not interact become dependent as a result of mechanical behavior. Towards that end, the proposed TRAUMAP system (Trauma Modeling of Anatomy and Physiology) consists of three modules: (1) a hypothesis generator for suggesting possible structural changes that result from the direct injuries sustained; (2) an information source for responding to operator querying about anatomical structures, physiological processes, and pathophysiological processes; and (3) a continuous system simulator for simulating and illustrating anatomical and physiological changes in three dimensions. Models that can capture such changes may serve as an infrastructure for more detailed modeling and benefit surgical planning, surgical training, and general medical education, enabling students to visualize better, in an interactive environment, certain basic anatomical and physiological dependencies. CCoommmmeennttss University of Pennsylvania Department of Computer and Information Science Technical Report No. MS- CIS-95-29. This technical report is available at ScholarlyCommons: https://repository.upenn.edu/cis_reports/197 TrauMAP Integrating Anatomical and Physiological Simulation (Ph.D. Dissertation Proposal) MS-CIS-95-29 Jonathan Kaye University of Pennsylvania School of Engineering and Applied Science Computer and Information Science Department Philadelphia, PA 19104-6389 INTEGRATIANNGA TOMICAL PHYSIOLOGICAL AND SIMULATION Jonathan Kaye [email protected] Computer and Information Science University of Pennsylvania 200 South 33rd Street Philadelphia, PA 19 104-6389 Bonnie L. Webber, Ph.D., Thesis Advisor John R. Clarke, M.D., Domain Consultant Abstract In trauma, many injuries impact anatomical structures, which may in turn affect physiological processes-not only those processes within the structure, but also ones occuring in physical proximity to them. Our goal with this research is to model mechanical interactions of different body systems and their impingement on underlying physiological processes. We are particularly concerned with pathological situations in which body system functions that normally do not interact become dependent as a result of mechanical behavior. Towards that end, the proposed TRAUMAP system (Trauma Modeling of Anatomy and Physiology) consists of three modules: (1) a hypothesis generator for suggesting possible structural changes that result from the direct injuries sustained; (2) an information source for responding to operator querying about anatomical structures, physiological processes, and pathophysiological processes; and (3) a continuous system simulator for simulating and illustrating anatomical and physiological changes in three dimensions. Models that can capture such changes may serve as an infrastructure for more detailed modeling and benefit surgical planning, surgical training, and general medical education, enabling students to visualize better, in an interactive environment, certain basic anatomical and physiological dependencies. Table of Contents 1. Introduction ...................................................................................1. . 2 . Objectives and Tasks ........................................................................4.. . ............................................................................. 2.1. Objective 6 2.2. Thesis Claim and Contributions ..................................................7. ...................................................................... 2.3. Proposed Work 10 ........................................... 2.3.1. Penetration Path Assessment 11 2.3.2. Static Anatomical and Physiological Knowledge ...................1. 2 ..................................................... 2.3.3. Dynamic Simulation 13 ............................................................................ 2.4. Summary -14 3 . Background ..................................................................................-.1. 5 3.1. Symbolic Anatomical Knowledge in A1 .........................................1.5 3.3. Anatomical Databases ............................................................. 18 3.4. Simulation in Medical Education ................................................2..2 3.4.1. The Road to Virtual Surgical Simulation .............................2. 3 3.4.2. Simulation in Medical Education .....................................-.2 5 3.5. Summary ...........................................................................2.8. 4 . Design and Method ..........................................................................-.2.9 ............................................................................. 4.1. Overview 30 4.2. Example Session ..................................................................3.1. ............................................................ 4.3. Component Information 35 4.3.1. Anatomical Object .....................................................3.6. 4.3.2. Physiological Process .................................................3.8. 4.3.3. Integration .............................................................3..8. 4.4. Theory and Implementation ......................................................3..9 4.4.1. Penetration Path Computations and Assessment ....................3. 9 4.4.2. Static Anatomical and Physiological Knowledge ...................4. 1 4.4.3. Dynamic Simulation ...................................................4..1 4.5. Preliminary Results ...............................................................5.6. 4.5.1. Penetration Path Assessment ..........................................5.6 4.5.2. Static Anatomical and Physiological Knowledge ....................5 6 4.5.3. Dynamic Simulation ....................................................5.7 ............................................................................ 4.6. Summary -62 5 . Conclusion ...................................................................................-.6.4 Appendix . Medical Terminology ...............................................................6.6. Acknowledgments ..............................................................................-..6 8 References .......................................................................................6.8.. 1. Introduction Anatomy is the study of the body structures. Physiology is the study of the essential and characteristic life processes, activities, and functions. Physiological processes are carried out within a physical space that may both influence and be influenced by the processes occurring within. This dissertation is about modeling anatomical structures, physiological processes, and some of their interdependencies as a result of mechanical interaction. Trauma frequently involves structural changes to the body, such as fractures, hemorrhage, and ruptured organs, which may affect multiple body systems. Interaction among body systems occurs not only due to functional dependencies but also due to constraints of the physical space they share. When the physical interconnectivity of body systems changes, we need to reconsider the dependencies among them to predict the effects of injury. Such assessment and resulting behavior will depend crucially on physical proximity and contact forces. For example, aflail chest is the condition in which segments of the ribcage become detached. During breathing, the flail section moves paradoxically to the rest of the ribcage because the net forces on that section are different from those acting on the intact sections. In a tension pneumothorax, accumulation of air within the intrapleural space results in pressing the mediastinum against the opposite lung. With the resulting increased pressure on the inferior vena cava in the mediastinum, the vein collapses and impedes venous return to the heart. With a little knowledge about cardiopulmonary anatomy and physiology, it is straightforward for us to predict the resulting behaviors. Part of this understanding depends on our ability to reason about the interaction of adjacent physical structures within an enclosed environment and the processes that change those structures. While we can quickly grasp the essential mechanisms that result in the behavior, we would be more hard- pressed to describe the particular effects in detail. Identifying the particular causes, effects, and associations, are critical for our understanding of physiological mechanisms, both in physiological research and teaching. The computer has the potential for elucidating that detail and presenting it in a visually- intuitive way for us, insofar as we can describe it. However, the computer does not implicitly share the insights we have about consequences of adjacency and physical change. Biomedical researchers have been using analog and digital computers to study physiological systems for some time now [18, 32, 37,421. Computer models are used as research tools for advancing physiological and clinical insight, models for indirect estimation of physiological parameters, models for control and therapy of on-line systems, and models for education and training [37]. Interest in biomedical research has grown recently in areas of Computer Science, specifically within the Artificial Intelligence community [49, 5 11. We are interested particularly in the computer's potential to impact medical education such as by simulating examples described above. The examples emphasize the critical relationship between the physical existence of an anatomical part with the functional role it plays. An accurate simulation of the body, then, requires an approach that integrates a realistic, three-dimensional structural model, deformable body dynamics, and physiological dynamics (mechanical, biochemical, and electrical)-a functional anatomy that explicitly links the anatomical structures with physiological behavior of the body. With new developments in computer hardware and graphics, and the success of flight simulators in pilot training, much talk has been made of creating virtual environments for medical training [19]. While people recognize the existing body of work in physiological modeling, we have not seen an application bridging that work with computer graphics modeling. We propose to develop a first principles approach for supporting 3-D visualization of anatomical and physiological interaction in the domain of penetrating trauma. We plan to accomplish this by integrating physically-based modeling methods for simulating anatomical parts with traditional physiological simulation. The TRAUMAP system (Trauma Modeling of Anatomy and Physiology) proposed consists of three modules: (1) a hypothesis generator for suggesting possible structural changes that result from the direct injuries sustained; (2) an information source for responding to operator querying about anatomical structures, physiological processes, and pathophysiological processes; and (3) a continuous system simulator for simulating and illustrating in three dimensions anatomical and physiological changes. We argue that this fundamental knowledge and presentation can be useful for illustrating medical concepts and conditions involving anatomical and physiological dependencies. We also see that such an approach ultimately may serve as an infrastructure for surgical planning and training. Outline of this Document We begin describing our research plan in Section 2 by describing the problems we face, the requirements for the project, the objectives of the study, and the specific tasks we propose to undertake. Section 3 reviews some approaches from the literature to anatomical and Introduction 2 physiological modeling. In Section 4, we outline our design and the methods we expect to employ, as well as detailing preliminary results. We conclude in Section 5 with a summary of critical themes. Introduction 2. Objectives and Tasks While the debate rages over exactly how many words a picture 'paints,' l it is obvious that pictures or diagrams can be effective in conveying certain concepts, particularly those which involve some reasoning about physical space. The computer affords a unique opportunity through visualization beyond static images for us to create and interact within virtual environments. We believe that a virtual environment can be useful to demonstrate concepts in addition to its obvious value for enabling interactive exploration of difficult-to- reproduce scenarios. The fact that the computer can present realistic looking images or animations does not imply that the computer has access to the knowledge of how the objects behave. In creating our virtual environment, we are concerned with linking physiological variables and parameters with anatomical structures in such a way that this combined object description can be applicable more generally, not just in specific, predetermined situations. Part of this involves developing methods to simulate physical laws so that the objects that populate the environment are exposed to and affected by them naturally. For example, a straightforward physical law dictates that objects cannot interpenetrate. If one object exerts a force on an adjacent object we need to model how that second object reacts to that force, its impact both on the object structure (anatomical deformation) and behavior (underlying physiological effect). In our virtual environment, we represent anatomical organs and physiological systems. Activity consists of anatomical, physiological, and pathophysiological changes. Characterizing these changes in a principled and generalizable way will be an important step toward fully-interactive simulation for education, such as environments for virtual surgery. At the core of a surgical simulator must be a modeling approach that captures accurate body system behaviors and presents them in a visually-plausible way. For the most part, the body is composed of fluids, soft tissue, and bones; such a core will require algorithms for modeling fluids, deformable objects, and rigid structures. Most importantly, it will require methods to manage their interaction. Before knowing how objects interact we need to know how individual objects behave. From this knowledge we may predict how changes from one object can affect another. There are at least two facets to object behavior, namely how the structure of the object can l~relirninarre~s earch indicates around 1,000. Objectives and Tasks change (based on its material properties and forces applied to it), and which processes, if any, are occuring within. In most cases, aspects of the processes occuring within depend on the physical object structure. For example, fluid flow through a vessel requires that no cross-sectional area of the vessel becomes too small to permit passage of the fluid. The problems faced in this pursuit are choosing, developing, and integrating methods for anatomical object and physiological process behaviors. We distinguish these facets by referring to the dynamics of physical shape and material modeling as anatomical modeling, whereas modeling the processes that cause the change we consider to be physiological modeling. More specifically, we focus our physiological modeling on the modeling of mechanical behavior, as opposed to related biochemical or electrical physiological processes. To model anatomical changes, we need to know such information as geometric descriptions of anatomical parts, relationships among them, material properties, and points of attachment. For physiological knowledge, we need to know such information as time-varying behavior descriptions, physiological variables, parameters, and associations among behaviors, conditions, and parts. Conventional physiological modeling describes systems as lumped-parameter or distributed-parameter models [5]. A lumped-parameter model, described in ordinary differential equations, expresses the cumulative effect over an element. In contrast, distrib~~ted-parametmero dels, described in partial differential equations, express the possible variation in individual segments of an element. For example, a lumped resistance would be a single value (or variable) representing the resistance encountered along the full length of an element. A distributed resistance would be an array of values representing individual effects of segments along the element. Such detail, however, may be more difficult (if at all possible) to observe clinically. Most physiological models of interest to us have been described in lumped-parameter form. This is consistent with clinical physiological instruction and observable physiological behavior (e.g., body surface pressure, chest wall volume change, etc.). The knowledge about relating physiological behaviors to anatomical parts involves knowing how physiological variables affect physical changes (to anatomical structures), and vice versa. For example, if a patient is lying with her lower extremities elevated (Trendelenburg position), as during a pelvic laparoscopy procedure, she may experience more difficulty breathing because the pressure from the abdominal contents exert more force against the diaphragm. Lastly, we need a mechanism to 'set the scene,' in other words to present the clinical data appropriate for the circumstance being modeled. For modeling penetrating injuries, we Objectives and Tasks 5