Dissertation Proposal

Modernization of the Mock Circulatory Loop

Advanced physical modeling, high performance hardware, and incorporation of anatomical geometry

Charles E. Taylor

12/1/2012

VCU BME: Artificial Heart Laboratory Dissertation Proposal Charles Edward Taylor

Proprietary and Confidential Abstract Page 2 of 25

Table of Contents

ABSTRACT ...................................................................................................................................................................... 3 PROJECT NARRATIVE ................................................................................................................................................... 4 FACILITIES AND OTHER RESOURCES ........................................................................................................................ 5

COMPUTATIONAL RESOURCES ......................................................................................................................................... 5 PROTOTYPING SUPPORT ................................................................................................................................................. 5 INVESTIGATOR SUPPORT ................................................................................................................................................. 6

EQUIPMENT .................................................................................................................................................................... 7 SPECIFIC AIMS ............................................................................................................................................................... 8 RESEARCH STRATEGY ................................................................................................................................................. 9

A SIGNIFICANCE ....................................................................................................................................................... 9 B INNOVATION ........................................................................................................................................................ 11 C APPROACH .......................................................................................................................................................... 12

Overview of the personnel and approach ............................................................................................................... 12 Preliminary studies supporting approach ............................................................................................................... 13 Aim 1: Develop computational plant models with process control logic. ................................................................ 15

Experimental Approach ........................................................................................................................................................ 16 Validation, evaluation, and benchmarks ............................................................................................................................... 17 Potential problems and alternate strategies ......................................................................................................................... 17

Aim 2: Deploy a fully automated mock circulatory loop .......................................................................................... 17 Experimental Approach ........................................................................................................................................................ 18 Validation, evaluation, and benchmarks ............................................................................................................................... 19 Potential problems and alternate strategies ......................................................................................................................... 19

Aim 3: Integrate critical anatomical geometries; left atrium and ascending aorta. ................................................. 19 Experimental Approach ........................................................................................................................................................ 19 Validation, evaluation, and benchmarks ............................................................................................................................... 20 Potential problems and alternate strategies ......................................................................................................................... 20

BIBLIOGRAPHY AND REFERENCES CITED .............................................................................................................. 21 RESOURCE SHARING PLAN ....................................................................................................................................... 25

DATA SHARING PLAN .................................................................................................................................................... 25 SOFTWARE SHARING PLAN ............................................................................................................................................ 25


Proprietary and Confidential Abstract Page 3 of 25

Abstract A systemic mock circulatory loop plays a pivotal role as the in vitro assessment tool for left heart medical

devices. The standard design employed by many research groups dates to the early 1970's, and lacks the acuity needed for the advanced device designs currently being explored. The necessity to update the architecture of this in vitro tool has become apparent as the historical design fails to deliver the performance needed to simulate conditions and events that have been clinically identified as challenges for future device designs. In order to appropriately deliver the testing solution needed, a comprehensive evaluation of the functionality demanded must be understood. The resulting system is a fully automated systemic mock circulatory loop, inclusive of anatomical geometries at critical flow sections, and accompanying software tools to execute precise investigations of cardiac device performance.

Delivering this complete testing solution will be achieved through three research aims: (1) Utilization of advanced physical modeling tools to develop a high fidelity computational model of the in vitro system. This model will enable control design of the logic that will govern the in vitro actuators, allow experimental settings to be evaluated prior to execution in the mock circulatory loop, and determination of system settings that replicate clinical patient data. (2) Deployment of a fully automated mock circulatory loop that allows for runtime control of all the settings needed to appropriately construct the conditions of interest. It is essential that the system is able to change set point on the fly; simulation of cardiovascular dynamics and event sequences require this functionality. The robustness of an automated system with incorporated closed loop control logic yields a mock circulatory loop with excellent reproducibility, which is essential for effective device evaluation. (3) Incorporating anatomical geometry at the critical device interfaces; ascending aorta and left atrium. These anatomies represent complex shapes; the flows present in these sections are complex and greatly affect device performance. Increasing the fidelity of the local flow fields at these interfaces delivers a more accurate representation of the device performance in vivo.

The expanded research capabilities of these combined tools will enable investigations of complex cardiovascular events and conditions previously unachievable. In addition to the findings and device design implications of those efforts, research in the biomedical engineering benefits from (a) the applied practice of using cutting edge physical modeling tools and control design methods for computational tool development of multi domain systems, (b) an in vitro testing solution that reflects advancements in cardiac device research, and (c) focusing a highly proprietary field towards established public databases.

This work is being developed under Gerald E. Miller, whose expertise and history in the field of cardiac devices will be an invaluable resource in the direction and dissemination of this work. Completion of work in his Artificial Heart Laboratory grants access to cutting edge research tools and facilities that will be crucial in the execution of the research aims of this proposed work.


Proprietary and Confidential Project Narrative Page 4 of 25

Project Narrative Automation of mock circulatory loop analysis increases the investigative capabilities of cardiac assist

technology assessment enabling clinical challenges that are currently not being simulated to be overcome. Producing an analogous in silico model enables the predetermination of mechanical settings that match clinical data, assisting this transition from the clinic to the bench testing environment. Incorporating anatomical geometries at the critical interfaces to the devices increases the acuity of the simulation, enabling more accurate assessments of performance and efficacy.


Proprietary and Confidential Facilities and Other Resources Page 5 of 25

Facilities and Other Resources Virginia Commonwealth University has provided an excellent site for development during the preliminary

investigations, and will continue to deliver the resources for this projects completion. The institution has extensive support for the electronics design, mechanical prototyping and software development necessary for the automated process control of the mock circulatory loop. The development space currently housing the preliminary designs is suitable for the project execution; floor space and facilities resources have been established. As part of this projects alternate strategy, external service providers have been identified and are presented to communicate the appropriate contingency planning contained in this proposal. Collaborations with key software and hardware vendors have allowed cutting edge products and technologies to be implemented in this current project. With this strong system of internal resources, external service vendors, and collaboration the execution of the current project is aptly supported through the Artificial Heart Laboratory at VCU.

Computational Resources

The physical modeling software validated in the preliminary designs and targeted for the project will be developed in The MathWorks (Natick, MA) Matlab and Simulink environments. A close collaboration with this company has allowed the laboratory to readily deploy some of the current products used in the construction of the computational models. As many of these products are cutting edge, and have minimal application precedent in biomedical engineering, the interest by The MathWorks to have successful installations is high. Support for these products has been well maintained, and the preliminary work has shown their success. Current product development, and future release timing, has been communicated by The MathWorks to ensure that this project takes advantage of the latest functionality.

Data archival and presentation tools are provided free of charge by Microsoft through the Microsoft Developer Network Academic Alliance (MSDNAA). These server and databases packages allow for the content generated by this project to be appropriately managed and accessed. The Artificial Heart Laboratory currently runs a SharePoint Server on a Windows Server platform with SQL Server database tools. This deployment allows for all content pertaining to this project to be web accessible with appropriate security measures in place for access restriction. It will serve as the repository that will be opened to collaborators, and certain content of this project made available to the public, per the data sharing plan.

The Biomedical Engineering department at VCU has research groups currently developing anatomical models for other projects. The proximity of these groups to this investigator allows for adequate technical support during the model construction process. Sharing of advancements in model construction techniques has been discussed, with support by the advanced users in those groups pledged to this project. With a large medical facility at VCU, there is a wide availability of consultants for assessing the anatomical features of the models.

The programming support within the engineering community at VCU provides sufficient resources for troubleshooting unforeseen software challenges. Drs. Bai and Pawluk of the Biomedical Engineering faculty provide programming expertise in close proximity to the research facilities of the Artificial Heart Laboratory. Assistance with the microcontroller programming and embedded electronics interfacing will be sourced from Zachary Dziczkowski, a graduate student in Computer Science and Engineering, with who a collaborative agreement is in place for this support. Further embedded electronics support can be solicited from faculty in the Computer Science and Engineering department, who reside in the two adjoining engineering buildings on the VCU campus.

Prototyping Support

The support of the School of Engineering at VCU allows access to a robust machine shop. All the fabrication in the preliminary designs was completed with this resource. Outsourcing of coatings necessary for the material compatibility has been orchestrated through the machine shop personnel and their industry contacts. Industry proven personnel administer the shop, and have been a valuable resource in design assessment. This laboratory has been a significant contributor to the Engineering School's machine shop, with instruction given by the investigator on cutting edge CNC programming products and fabrication of anatomical structures. The machine shop capabilities and information sharing will continue to be an integral part to the success in the mechanical systems development in this project. In the event that a design cannot be


Proprietary and Confidential Facilities and Other Resources Page 6 of 25

manufactured with this facility, Khem Precision Machining (Richmond, VA) will be contracted to fabricate the parts.

The Artificial Heart Laboratory has developed its electronics prototyping resources to the extent that it can fabricate its own circuit boards in house from raw materials. The capabilities allow for trace isolation on copper clad boards, tinning, surface mount reflow, hot air rework, and soldering. These functions enable circuit boards to be produced for a fraction of the price external resources charge. In the event that there is a design that cannot be manufactured with this equipment, Advanced Circuits (Aurora, CO) will be used as a contract service to supply the boards.

Support for the electronics design is available through the faculty in the Biomedical Engineering department. Drs. Wetzel and Fei provide tremendous expertise when consultation on electrical circuit design is required. Their combined knowledge of high and low wattage designs will be useful during this projects work on actuation and sensor signal conditioning, respectively. Outside of these intradepartmental resources, the department of Electrical Engineering's faculty has many individuals that can provide consultation on the issues this project may encounter.

Investigator Support

The investigator for this project is currently funded as a Research Assistant, which allows the full dedication of his time to this project. There are no major obligations outside of the work contained in this proposal. Minor obligations include Laboratory Manager duties with respect to safety inspections and personnel management. Currently, there are no undergraduate researchers directly reporting to this investigator. A second year doctoral student is on hand to assist with some of the aims of this proposal.

Training: Skill development has been supported by the Artificial Heart Laboratory in the form of

product training courses that have been attended by the investigator. Attendance of future training courses should not be necessary at this point in time. Support to attend application training can be arranged if necessary. Departmental interaction with the aforementioned faculty supporting the different facets of this project also enables cross-training to occur. Coursework and lecture series complement the educational resources available for research skill development.

Mentoring: The mentor to this investigator, Dr. Miller, has an established record of cardiovascular research. His expertise in the area of mock circulatory loops and medical devices provide an invaluable resource for this investigator. Consultation on system features and research field gaps will aid in positioning this projects work as a cutting edge simulation platform for cardiovascular device assessment. Having good rapport with most of the Biomedical Engineering faculty, this investigator has a broad base for career development counseling and leadership advice.

Facilities and Administration Support: The facilities of the Artificial Heart Laboratory have been sufficiently outfitted, with no foreseen issues. Administrative support through the Biomedical Engineering department has been exceptional. Supplies purchases have been streamlined to allow for next day delivery to be achieved from preferred vendors. The research to date in this laboratory has benefited greatly from this administrative support.

Development and support resources available to this investigator through the Biomedical Engineering

department at VCU are extraordinary. Their presence will be invaluable through the execution of this projects aims, and will be integral to the development of this investigators research capability.


Proprietary and Confidential Equipment Page 7 of 25

Equipment As discussed in the Facilities section, the resources available to this investigator at VCU are exceptional.

Much of the work contained in this proposal is to be developed from raw materials, integrated circuit components, and code generation tools. The enumeration of the critical hardware and software packages used to complete the aims of this proposal are presented as follows:

Computational resources: Two 3.0 GHz Core 2-Duo Windows XP 32-bit machines with 4GB of RAM, 240GB hard drives and Radeon workstation graphics cards. The two stations are used for software programming and CAD work. Five Pentium 4 Windows XP 32-bit machines with 4GB of RAM. These systems are used for data acquisition, prototyping, and data servers. One Dell T2200 workstation with Quad Core 3.0GHz i7 Pentium Windows XP 32-bit, 4GB of RAM and Nvidia Quadro 2000 with CUDA 2.0 GPU support. This system is tasked with PIV control, PIV processing, and GPU processing.

Software Resources: SolidWorks Professional licenses that include the Motion, Simulation, and Flow Simulation packages. MasterCAM for SolidWorks provides the machine code generation for the CNC work. The MathWorks toolboxes central to the completion of this proposed work are: Simulink Coder, Matlab Coder, Embedded Coder, Control Design Toolbox, Parameter Estimation Toolbox, Optimization Toolbox, and Fixed Point Toolbox. National Instruments LabVIEW with the Real-Time Module allows for robust data acquisition solutions.

Electronics: Electronics prototyping equipment in the form of an Agilent 10MHz oscilloscope and an HP signal generator are currently owned by the laboratory. National Instruments PXI hardware with embedded controller and data acquisition hardware allow for accurate data sampling and real-time signal processing to be achieved. An assortment of integrated circuits, passive components, and microcontrollers has been compiled in the laboratory to enable embedded electronics research to be appropriately supported. High performance pressure sensors (Entran EPX model) and accurate flow meters (Transonic T100) utilized by the laboratory ensure that effective measurement technologies are employed for the in vitro system.

Fabrication resources: The laboratory has acquired a T-Tech PCB mill that is capable of isolating surface mount footprints and comes with isolation routing software. A Haas 3-axis CNC, Stratasys FDM rapid prototyping machine, lathe, Prototrac enabled milling station, welding station, and an assortment of tooling yields a well endowed machine shop available 5 days a week through the school of engineering. Additional usage can be contracted through the school of engineering for $20/hour.


Proprietary and Confidential Specific Aims Page 8 of 25

Specific Aims Mock circulatory loops are employed in ventricular assist device, total artificial hearts, and prosthetic valve

evaluation as in vitro analogues of the cardiovascular system. These systems function as both basic research tools for design development and preclinical evaluation units for safety verification. The design of the mock circulatory loops in use now has not deviated much from their initial embodiments of the 1970's; fluid tanks with variable air spaces and manual valves. As the technology of the implantable devices has seen major innovations, the in vitro assessment tools should reflect this development and provide a more advanced testing solution.

This project proposes a high performance mock circulatory loop, inclusive of critical anatomical geometries, and accompanying computational tools for a state of the art solution to cardiac device assessment. A combination of novel hardware and application of industrial software solutions will be used to produce a system with a level of functionality that is currently not available in the field, allowing for previously unachievable simulations to be performed. The knowledge of the team in the areas of device prototyping, embedded system programming, and medical device research ensure the deliverables of this project will be realized through the following aims:

Aim 1: Develop computational plant models of the mock circulatory loop inclusive of process

control logic Challenge: Develop an in silico physical system model that is capable of accurately simulating an

analogous in vitro platform; it must also include the monitoring algorithms and control logic governing the actuators.

Approach: Utilize commercially available physical modeling tools to rapidly develop the computational models. Parameter estimation functions and control design tools will be used to tune the model components to experimental performance data and develop the appropriate controller methods, respectively.

Impact: Utilization of industrial engineering tools in developing these computational resources, and executing the control design, illustrates an application of this software technique not widely applied in this field.

Aim 2: Deploy a fully automated mock circulatory loop Challenge: Fabricate an in vitro embodiment of the computational model that represents the performance

bandwidth and control acuity seen in silico. Approach: Construct the following subsystems of the mock circulatory loop: ventricular simulator, arterial

compliance chamber, peripheral resistor, and pulmonary simulator. Utilize commercial code translation products to deploy the computational control elements onto the embedded control systems hardware.

Impact: This system will provide a robust testing platform with functionality that is currently not available in this field; nonlinear arterial compliance and change of all controlled set points in runtime.

Aim 3: Integrate critical anatomical geometries; left atrium and ascending aortic root. Challenge: Incorporate the anatomical geometry of the left atrium and the ascending aortic root, which

need to be based off a publicly available patent data set. Approach: Extract the geometry from the Visible Human Project, process the results into a model capable

of fabrication, engineer the interfaces to the mock circulatory loop, and produce the in vitro pieces. Impact: Inclusion of these anatomically correct sections develops the local flow field characteristics that

have been shown to have significant impact on device performance. This integrated multi scale research tool approach provides an investigator with the (a) in silico tools that

allow for effective experimental design, (b) in vitro performance that enables a wide range of cardiovascular parameters to be simulated, and (c) accurate local flow conditions in the regions of the devices being evaluated. These deliverables provide the functionality that has been in increasing demand as the need for experimental acuity has risen.

Principle Personnel Charles Taylor, VCU Artificial Heart Laboratory: Design, development and deployment of the mock

circulatory system and computational tools. Gerald Miller, VCU Artificial Heart Laboratory: Advisor of project design goals and consultant on applied research stemming from the deliverables.


Proprietary and Confidential Research Strategy Page 9 of 25

Research Strategy A Significance

Translational medicine and human health impact are the hallmarks of funding packages granted through the NIH. Advanced in vitro methods for left ventricular assist device and prosthetic valve assessments have been identified as catalysts for more successful device designs, resulting in better clinical outcomes [1]. Research groups that have implantable device projects employ mock circulatory loops as their means for evaluation, but mainly rely on traditional designs. Steady state cardiovascular conditions and highly idealized simulation of cardiovascular system function has been the mainstay of device assessment in these traditional systems [2]. Clinical studies have shown that current limitations in quality of life with implanted patients arise from changes in cardiovascular conditions (e.g. postural changes, sudden exertion) [3]. It is necessary to test the emerging technologies against the challenging conditions witnessed in the clinic in order to address these issues.

Translating these physiological conditions into an early stage in vitro environment necessitates a reflective analog of the cardiovascular system. This analog must have the functional capacity to have runtime parameter control, recursiveness, and flow field similarity at the interfaces to the devices being tested. Although the cardiovascular system can be functionally represented in a lumped parameter model, embodied as a mock circulatory loop, the capability of executing changes in the parameters without stopping the system is key for constructing complex events [4]. Additionally, simulating dynamics enables the analysis at a multitude of stationary settings in a battery testing method; this could screen a device against a large set of conditions each representative of a target patient population. The second point of functionality, recursiveness, is integral to the closed loop circulation of the cardiovascular system. Designing the appropriate system to correctly provide the recursive response can be challenging; it has many dependencies and is highly dynamic. Inclusion of anatomical geometries at interfaces of the mock circulatory loop and the devices being tested is crucial. The accuracy of these local flow fields have been shown to have a strong effect on heart valve function and pump efficiencies[5,6]. The investigator of this proposal has made advances in the deployment of a mock circulatory loop design, through the revision of critical subsystems, which address the functionality necessitated for an in vitro cardiovascular system model. Continued efforts on the remaining elements of this model are crucial to providing this field with the appropriate investigational capability.

Robust in vitro testing has been identified by regulatory agencies, and governing scientific bodies, as integral to the safety assessment of devices in this field [7]. This investigator has identified key software designs and completed hardware developments that satisfy these agency initiatives and provide the requisite testing solution. Experimental condition design tools based on software packages being used for system modeling are readily adaptable to satisfying this proposed work. Use of PhysioBank patient databases as in vitro condition sources connects this work with current clinical efforts, and can only be achieved with updating the current mock circulatory loop technology [8]. Applying Visible Human Project data as the basis for anatomical models continues the effort to utilize pre-existing government databases, which enables the comparative analysis by other researchers [9]. The strategies chosen by this investigator highlight the value of using currently funded databases and commercial engineering tools to produce high functioning investigational tools and establish design paths for continued development in this field.

Applying these strategies to the development of a new mock circulatory loop model addresses the limitations of presently implemented systems. The restrictions of current models fall primarily in the insufficient automation and anatomical irrelevancy of the systems. This investigator proposes that developing a robust experimental system, with supporting computational tools, would advance research in this field through the following:

Illustrate the application of sophisticated physical modeling software in the construction of

mock circulatory loop computational models (Aim 1). Computational models have been developed concurrently with mock circulatory loops by other groups [10,11]. However, the acuity of these models has not provided the fidelity that is required for constructing precision control systems or design of experiments with dynamic conditions. The construction of a model through a method that is not based on hydraulic first principles (e.g. adaptation of electrical modeling packages) is typically the culprit of poor performance [12]. Physical modeling packages that are built for the systems they are representing (hydraulic, mechanical, etc.) produce exceptional performance in these models [13,14]. The goal of translating the application of these modeling methods into



biomedical engineering, specifically the medical device field, has large ramifications in the design efficiency by adopters of these tools.

Application of embedded control techniques to solve current limitations in mock circulatory

loop performance (Aim 2). Control of particular mock circulatory loop parameters has not been successfully implemented by other groups in this field. Attempts at controlling arterial compliance with traditional sensors and hardware have not been effective, until implementation of a novel control strategy was deployed by this investigator [15]. Inertial effects within the system are another area of difficulty for many designs, especially when trying to correctly model the cardiovascular recursiveness [16]. The commercial applications utilized in efficiently constructing the embedded control systems and novel hardware has been of particular interest to the in vitro groups of this field.

Free anatomical CAD model development path with low cost fabrication method (Aim 3).

Development of anatomical structures for investigations is a resource intensive task; dataset acquisition, cost of modeling software, and cost of model fabrication [17]. This proposal seeks to illustrate a method for utilizing the Visible Human Project data in the construction of anatomical CAD models through open source software. The fabrication method chosen for these models requires resources typically available to engineering groups; a CNC mill as opposed to stereo lithography (SLA) systems. The basis of this work on a publicly available dataset seeks to cut the cost of model development and homogenize the models used by research groups.

Automated mock circulatory loop, including anatomical sections, with accompanying

experimental design software for accurate investigation of cardiac devices. Supplying this field with a robust and accurate testing system enables many of the pressing clinical concerns to be effectively investigated. Advancing the state of preclinical evaluation tools will have a sizeable impact on the investigational quality and technological advancement of the devices developed with this platform. This project not only proposes this completed system, but publication of the workflows (Figure C.1) applied in the achievement of this objective may enhance research capabilities in this field through the following:

Aim 1: Modeling process for constructing the plant model of the mock circulatory loop can be extended to device computational models. Ventricular assist device and total artificial heart developers working with multi-domain modeling will greatly benefit from the deliverables of this project. Providing this field with these cutting edge methods accelerates development efforts on other funded projects currently burdened with the computational modeling of these complex systems. Aim 2: Embedded control systems have an increasing role in device development in this field. Introducing development tools that assist in the translation of complex control architecture onto the target hardware is beneficial to many groups still utilizing manual coding techniques. Implementing code generation tools that were developed for regulated environments addresses future compliance issues with the FDA in this area [18]. Increasing the code deployment throughput synergizes the research return for the device projects and streamlines the development process.

Figure C.1: Diagram of the objectives and how the work is encompassed by the respective aims. In silico work is grouped in the upper half of the illustration, with the in vitro analogues for each presented in the lower half. Computational model work on the subsystems representing the elements of the mock circulatory loop is delivered by Aim 1 (red outline). Fabrication of the hardware and incorporating the subsystems into a mock circulatory loop is presented in Aim 2 (blue outline). The development of the anatomical CAD models and production of the in vitro pieces is presented in Aim 3 (green outline).



Aim 3: Anatomical geometry inclusion is crucial to many biomedical engineering projects. The use of cryoslice data in this project and open source software are novel, with a broad spectrum of applications in this field. The promotion of this data source, and software packages, will hopefully enable research that was previously cost prohibitive without these resources. This proposal promotes the solutions an automated mock circulatory loop will provide, but it also illustrates multiple design tools that have significant implications on applied biomedical research. This modernization of design tools is crucial to successfully implementing high fidelity testing tools and technological development of devices in this field.

Public dissemination of methods, architecture, and resulting data. The utilization of public

datasets for both the anatomical geometry (Visible Human Project) and conditions (PhysioBank) were deliberately chosen so that other researchers could apply the work of this proposal more easily. In following this open source data model, contact has been made with the administrators of both databases to procure resources for data publication (see letters of support). Publishing the simulated event data (in silico and in vitro) in the PhysioBank, delivers on one of the missions of this database being a repository for cardiovascular conditions data. Submission of the anatomical CAD files to the Visible Human Project seeks to establish a forum in this database for geometries that are based on the imaging data. These data sharing initiatives are crucial to this investigators vision of common data sources in this field and higher utilization of previously funded repositories.

The deployment of the automated mock circulatory loop, and inclusion of the anatomical sections, is crucial

to the advancement of this platform as a modern approach to device assessment. The broader implication of this work lies in the methods and data sources employed. Accelerating development with these approaches has the ability to impact a wide range of research. Enabling this research base with these cutting edge methods has the potential to increase the momentum in other funded projects and accelerate progress towards the public health solutions they are pursuing.

B Innovation

With the functionality of current mock circulatory loop designs failing to successfully translate the clinical performance challenges to the in vitro setting, a modernization of this testing solution was needed. Applying advanced physical modeling software to this field proved to be a novel approach in satisfying the high fidelity requirement, and quick development timeline, for a computational model. This modeling approach allowed for model elements, based on their respective first principles, to be integrated across physical domains and incorporated with the control logic. Utilizing embedded code generation tools, the computational models control logic sections were implemented on the embedded processors of the mock circulatory loop. This seamless model-to-code approach is a cutting edge technique translated to this area of research [19,20].

Automation principles applied to the mock circulatory loop have also proven to be novel for this field. Measurement and control of all runtime parameters has resulted in novel functionality in this design. Control of arterial compliance in runtime was found to be a first for this research field, and enables a large range of simulation capabilities previously seen as unobtainable [21]. The success with arterial compliance is exemplary of the advances a total automation perspective brings to the mock circulatory loop; cardiovascular dynamics simulations can be conducted and nonlinear compliance can be modeled.

Implementation of public databases for both the conditions design and anatomical geometry is an innovative means of homogenizing the basis for device evaluation. Data mining the PhysioBank resources for challenging and exemplary conditions delivers a basis for the resulting experimental design [22]. This resource continues to offer new datasets, which feeds the experimental capacity of the investigational tools connected to it. Taking advantage of the cryoslice data in the Visible Human Project delivers a unique solution to the high resolution requirements of modeling soft tissue. Creating a contributable database for extracted geometry based on this dataset expands the offering this source provides. Implementing these models in a physical form within the automated mock circulatory loop builds on the experimental data that is attributable to these open source geometries; this assists other research groups by providing validation datasets to their in silico results based on this anatomy.



C Approach Overview of the personnel and approach

The production of an automated mock circulatory loop was postulated during previous investigations where gaps present in the existing testing solutions limited the fidelity of the simulations[2325]. The aims of this proposal were constructed from the identified performance requirements needed to satisfy current ISO certifications for medical devices and desired features that would meet future demands of this system [2628]. Datasets obtained in one aim may initiate work in another, resulting in a staggered progression through the stated objectives (Figure C.2). Although innovation tasks are reserved for the PI, support work during the implementation process will be partially subsidized by the staff of the Artificial Heart Laboratory and its collaborators. The dissection of the personnel and approach is discussed here.

Project leadership and system development: The PI, Charles Taylor, will be the project manager for the work contained in this proposal. His previous industry experience in designing high throughput ADME profiling systems has endowed him with the technological experience and design management skills needed for successful completion of this work [2931]. Application of his knowledge and leadership has resulted in successful research projects in the Artificial Heart Laboratory, 3 primary authorships of peer-reviewed journal articles on advancements made to mock circulatory loop design, and 14 conference presentations on material supporting this proposed work. The success of this project lies in Mr. Taylor's expertise in automated testing systems and the management skills he will apply in its execution.

Programming work: The PI will be responsible for development of code for this project. Pre-implementation debugging and testing routines might be delegated to lab support personnel to enable the investigator to pursue other objectives of this project. Training on the embedded code generation tools was identified by the PI as necessary for successful implementation of those methods; education on the software, with no consultancy on the application, was provided through commercial training resources. Collaboration with a Computer Science investigator has been established for embedded hardware programming support.

System fabrication work: Production of the system components will be performed by the PI. His experience in custom fabrication and prototyping techniques enables these components to be produced for a fraction of the outsourced price. Utilizing CAD/CAM packages and having prior knowledge of CNC systems, the PI will be able to effectively produce the elements of the system that would be challenging and costly to contract for production (e.g. anatomical sections). In-laboratory development of electronic prototypes has accelerated the production of embedded control solutions, and positions this lab as a desirable collaborator within the institution. Coordination of fabrication resources (e.g. machine shop rental) and training of Artificial Heart Laboratory personnel on fabrication methods will assist the production of the components needed for this system.

Anatomical model work: Development of the anatomical models will be coordinated by the PI and executed as part of a team effort by the members of the Artificial Heart Laboratory. As this is a new capability in this laboratory, it is the interest of all personnel to be involved in this project aim. The PI will make the decision on the software package to use, once all options have been explored. The guidance on the geometry extraction and model construction practices will be of his design. Preparation of the model for fabrication will be the responsibility of the PI, due to his background in prototype construction.

Timeline: Preliminary mock circulatory loop design encompassed by Aim 1 will be pursued first. Prototypes and characterization of the mock circulatory loop subsystems outlined Aim 2 will be developed next. Design of control logic and full computational models will be completed in the next phase as part of Aim 1. The completion of Aim 2 will result in the application of the control designs and deployment of the mock circulatory loop. Aim 3 will be pursued at intervals through previous phases, with targeted implementation of the models coming after completion of the automated mock circulatory loop. The final stage is producing datasets from the system, disseminating the work, and coordinating database contributions.

Figure C.2: Gantt chart illustrating the dependencies of this project and the relative timeline.



Preliminary studies supporting approach Groundwork for this project proposal was initiated to test the viability of the development techniques and to

provide proofs of concept for certain hardware designs. The discussion of these preliminary research results are presented as case studies and illustrate how the aims of this proposal will be completed.

Peripheral resistor design and characterization: The method of applying a physical modeling package to the characterization of a piece of mock circulatory loop hardware was recently published [32]. This method illustrated the application of parameter estimation techniques in deducing the constitutive equation values for a custom device from experimental data. This method was applied to a flow control valve with a variable area orifice as the control mechanism. A custom design was pursued due to large operating bandwidth and automated precision needed for the intended investigations; the operating requirements were not satisfied with commercially available products [33]. The ability to elucidate the discharge coefficient and critical Reynolds number assisted in refining the accuracy of the valve's computational model [34,35]. The MathWorks software environment was chosen as the modeling platform due to the well supported physical modeling libraries and parameter estimation tools. The computational model for the valve was constructed with the Simscape physical modeling toolbox using the pre-existing hydraulic library blocks. The Parameter Estimation Toolbox's least squares algorithm was used for the determination of the computational model values by fitting to experimental data. The work on characterizing the peripheral resistor was first in the development process as this was seen as the easiest mechanism to model and test the parameter estimation process. The success of this method illustrates a workflow capable of developing high fidelity computational models from in vitro components, which is a critical aim of this proposal.

Automation of the Harvard Apparatus Pulsatile Blood Pump: The ability to implement embedded control on a subsystem using logic developed in a computational model was validated and published as the work on the automation of the Harvard Apparatus Pulsatile Blood Pump, with the finished product shown in Figure C.3 [36]. Utilizing the methods employed on the previously mentioned peripheral resistor work, a complete computational plant model of the Harvard pump was developed. This was constructed via parameter estimation from experimental data collected in studies that targeted characterization of the different domains of this pump; electrical, mechanical, and hydraulic. The driveline of the system was modeled in SolidWorks and imported as a rigid body system into Simulink via the SimMechanics Link [37,38]. This enabled an accurate depiction of the driveline to be represented without the investment of time to develop the equations for the linkage dynamics. Interfacing the multi-domain aspects of this model was also proven with the electromechanical motor to driveline to hydraulic pump head network.

With this accurate computational representation of the system, an accurately tuned algorithm was developed for process control of this pump. Control design and optimization tools were employed to build and tune the logic governing the pump operation. Ultimately, a feed-forward PI control topology was implemented to produce exceptional pump performance. This resulted in a pumping mechanism that greatly exceeds the manufacturers stated performance and effectively maintains set points for pumping values (Figure C.6).

Figure C.3: Annotated picture of modified Harvard Apparatus Pulsatile Blood Pump: manual stroke volume knob (1), manual pumping controls (2), manual/auto switch (3), diastole/systole out (4), USB port (5). Figure C.4: Illustration of percent systole change and control.



Development of the communication scheme that would be applied to the other subsystems in the mock circulatory loop was part of this preliminary work. The use of USB interfaces for the microcontroller communication yields a modern connectivity approach that is easily scalable for future designs; large communication bandwidth and ubiquitous connector interface. The communication drivers that were developed for this pump controller were designed to be translated easily to other subsystems.

The efforts in this published work embody the perspective of this proposal's strategy and the workflows designed to achieve its aims. The methods employed are cutting edge in physical plant modeling and control design. The success of this formative work substantiates these methods as the appropriate path for achieving the performance necessitated and efficient investment of effort by the investigator.

Implementation of a Novel Compliance Control Chamber: The real-time control of arterial compliance had not been effectively achieved in this field. Previous attempts have been based on open loop control of the arterial compliance with no established means of changing the compliance and effectively measuring the new setting [15,39,40]. Establishing this capability was integral to the targeted simulation of cardiovascular dynamics where the arterial compliance plays a role in severe pathological events and injury states [21]. This functionality also enables battery testing devices through a multitude of conditions without stopping the flow of the system.

Building on the principals of the peripheral resistor and Harvard pump automation, work on a novel arterial compliance chamber design (Figure C.5) with accompanying controller was completed [41]. Development and control of this device employed the same system characterization through parameter estimation, computational model construction, and control design methods previously discussed with the peripheral resistor. The novelty of this design was the application of well known membrane solid mechanics [4244]. Utilizing a laser displacement sensor with 3 micron resolution, the center deflection of a highly distensible circular membrane could be accurately resolved into the volume displacement. Using this membrane as a barrier between the hydraulic system of the mock circulatory loop and a closed pneumatic chamber, the volume expansion of the mock loop could be accurately controlled. This control is achieved via air pressure regulation of the pneumatic chamber that consequently governs the expansion of the membrane for a particular hydraulic pressure. Utilizing the computational model for this device, a process controller was developed with exceptional results of runtime arterial compliance measurement and set point change (Figure C.6). A novel aspect of the control design is the capability to maintain compliance on a set pressure-volume curve. This enables investigations of nonlinear compliance effects in vitro; a well-known tissue mechanics fact with no mock circulatory loop application to date [45].

The design deployed illustrates this investigators capability of developing novel solutions to obtain previously unachieved testing performance. The success of the work entailed in this publication falls on the workflow illustrated in the previous preliminary work and its capability of enabling rapid development of novel

Figure C.5: Compliance chamber design: pneumatic lines (1), displacement sensor (2), pneumatic pressure sensor (3), bleed port (4), hydraulic pressure sensor (5), hydraulic chamber (6), membrane (7), pneumatic chamber (8), fastening hardware (9), rail mount (10).

Figure C.6: Runtime compliance change and control illustration. The initial discrepancy in the computational model is due to the initial conditions prescribed.



solutions. With this critical investigation of arterial compliance control methods complete, the aim of this proposal in producing a robust mock circulatory system is further supported.

Fabrication of the ascending aortic root from Visible Human Project Cryoslice data: Anatomical sections that interface with the devices being assessed in the mock circulatory loop must be accurately depicted, as their geometry has considerable impact on the local flow fields governing device performance [46,47]. Inclusion of these geometries in mock circulatory loop designs has become standard practice, with both rigid and flexible models being included by various groups [48,49]. Incorporating these anatomical features into the flow path of the current mock circulatory loop has been the subject of a continuing line of presentation spearheaded by this investigator.

The unique approach to development of these models has been the open source nature of the model construction process. Utilizing a public image dataset, and constructing the model with free software, enables the results generated by this group to be more easily compared to other conclusions reached from this public data. The lack of publicly available high resolution MRI or CT data resulted in a novel approach of using cryo-slice data to extract the aortic geometry. The Visible Human Project, through the National Library of Medicine, provides a comprehensive set of cryo-slice images with square voxel dimensions of 0.3 mm. ITK-SNAP[50] was used to extract the aortic lumen boundary and construct an STL file of the results (Figure C.7A). Processing of this STL file was achieved through the use of Blender for smoothing and facet reduction [51,52]. The model was imported into SolidWorks to construct a lofted model (Figure C.7B), which facilitates the fabrication [53].

The ascending aorta was initially constructed to test the feasibility of prototype manufacturing through subtractive fabrication. This method was investigated for its ability to produce PIV capable prototypes with excellent optical clarity, which can be an issue with 3D printing methods(e.g. SLA) [17] . This approach is also a commercial technique for producing complex geometries in molds, which would become useful for distensible models that would need to be cast [54]. The success of this method was realized with an ascending aorta section machined from acrylic (Figure C.7C), and connected to the mock circulatory loop, to evaluate a tilting disc valve (Figure C.7D) [55]. Completion of this preliminary prototype concluded a large effort to develop an anatomical model construction process that maintained high spatial resolution and that was open source based. The latter continues the efforts of this PI in connecting the work of this proposal with other funded initiatives, and basing research results on data that is publicly available.

Aim 1: Develop computational plant models with process control logic.

In order to appropriately utilize the potential of the mock circulatory loop, a computational model of the entire system will be constructed using physical modeling software (see Figure C.8). This computational model will be used with parameter estimation tools to determine system settings that reproduce chosen clinical blood pressure waveforms.

The preliminary work describes a well formed technique of applying the Simscape physical modeling tools to the computational modeling efforts of this proposal. Expanding this work to encompass the entire mock circulatory loop, with governing logic, will present modeling challenges not previously encountered. The preliminary work was subsystem focused, with the initial conditions and boundary conditions being specific to those isolated investigations. Adaptation of these subsystem models will need to be performed, as they cannot be simply connected.

A B C D

Figure C.7: Stages of the aortic root model: ITK-SNAP trace (A), lofted model (B), machined prototype with particle flow (C), PIV data from tilting disc valve study using prototype (D).



Experimental Approach Computational modeling techniques typically employed for modeling mock circulatory loops have involved

the application of electrical analogs in the form of Windkessel models [11,56,57]. This method of modeling makes large assumptions when used for hydraulic systems and is limited when incorporating active controls for system elements. The utilization of physical modeling packages based on the first principles that are being modeled enables the computational model to simulate the nuisances of those particular domains. Effects of local resistances, discharge coefficients, and connection height differences can be incorporated using these physical modeling tools. The perspective of these modeling packages is such that the computational model should be reflective of the in vitro system; whereas the Windkessel approach models the cardiovascular system and the mock circulatory loop is an approximation to the in silico model. The governing principle of the modeling approach taken here is that the performance of the computational model is its ability to replicate physiological waveforms and match the in vitro behavior. Accurately modeling the mechanisms of the in vitro system satisfies the latter, with the former being delivered by appropriate design of those mechanisms to deliver the waveforms desired.

1. Constructing a venous reservoir model completes the needed elements of a systemic mock circulatory loop typically employed [58]. The reservoir design involves a decoupled return line to prevent perturbations from being delivered to the pump. This type of hydraulic design is not reflected in the Simscape libraries, which requires that this particular modeling element be custom programmed [59]. This will involve identifying the constitutive equations for a hydraulic tank with a free surface that is not pressure controlled. The governing equations will be modified to account for the flow contribution of the return line, but not the pressure. Coding this custom hydraulic element is possible, and will produce a Simscape modeling element that is capable of being incorporated with the rest of the physical model.

2. Design a pulmonary simulator that allows for appropriate simulation of the supply to the left atrium. This model should include a mechanism for pressure development (e.g. pump model) and control of the pressure as a function of flow delivered[60]. The inlet to this simulator will be the venous reservoir tank and the outlet will be connected to the Harvard pump inlet. Model elements of this subsystem should be based on real hardware performance values, to assist in the translation of this system to the in vitro setting.

3. Calculating the initial conditions to connect the subsystems will involve the identification of the steady state no flow operating point of the system, and correcting the subsystem parameters. This will involve applying hydraulic principles of column height and utilization of the membrane deformation data from the arterial compliance chamber investigations. A steady state low flow (~1 lpm) initial condition will also be determined to assist the model execution time, as starting the model from a no flow state comes at a large computational cost due to the inertial terms in the hydraulic elements. The accurate determination of these initial conditions is crucial to ensuring the model will execute with the various subsystems connected, as well as expediting the initialization computations.

4. Construct a full mock circulatory loop model from the subsystems discussed in the preliminary data and the venous reservoir model. This comprehensive model should couple the performance of the validated subsystem models and include simulated sensor data from points in the mock circulatory [61]. This embodiment strives to reproduce the interaction of all the elements present in the in vitro system; all physical domains and interfacing elements. This becomes the full model standard from which computational optimizations can be made.

5. Accelerate execution speed of model through the use of code packaging tools, solver choices, and simulation methods. Optimization tools exist to reduce code execution time by eliminating redundancies and developing transfer functions for eligible code segments[62,63]. Choice of numerical methods for solving the

Figure C.8: Simulink model of the mock circulatory loop. System settings are imported from the workspace and parsed to their respective subsystem controllers (upper left). The controller blocks (green with drop shadow) represent the logic governing the performance of the connected physical models.

Inputs

SetPointsOuputs

ResistanceController

PS Controls PS Sensors

PS Inlet PS Outlet

Pulmonary Simulator [PS]

PR Controls PR Sensors

PR Inlet PR Outlet

Peripheral Resistor [PR]

Inputs

SetPointsOuputs

PS Controller

HP Controls HP Sensors

HP Inlet HP Outlet

Harvard Pump [HP]

Inputs

SetPointsOuputs

HP Controller

Inputs

SetPointsOuputs

Compliance Controller

CC Controls CC Sensors

CC Inlet CC Outlet

Compliance Chamber [CC]

1SetPoint File

Pulmonary Simulator SetPointsHarvard Pump SetPointsCompliance SetPointsResistance SetPoints



system of equations associated with the physical modeling networks has a large impact on simulation time. Tuning the rules governing the solution convergence and step size can produce performance benefits. Utilizing a fixed step solver can be tested to see if there is a performance gain by reducing small steps taken by variable step solvers. Utilization of local solvers for the physical networks, and a discrete execution for the logic blocks, has the best acceleration capability for this model. An xPC target approach can also be investigated to deliver a real time model execution rate [64]. The need for solvers in the physical modeling networks limits the acceleration capabilities of the total model; software tools to expedite the solver process are limited.

6. Develop parameter estimation tools for settings determination using the total mock circulatory loop model. The computational model should be tuned to match a physiological pressure waveform, providing the settings needed for reproduction of that data in vitro. The source of the waveforms to be matched has been identified as the following PhysioBank databases containing blood pressure data: MIMIC, MIMIC II, and MGH/MF [8,65]. A graphical user interface will be developed to identify the waveform of interest in the database, apply parameter estimation of the model to determine the settings needed to replicate that waveform, and export those settings to a file that can be used as a simulation list for the mock circulatory loop. Connectivity to the PhysioBank ATM interface will be explored, as this interface is easy to use and reprogramming this functionality is not efficient. The Parameter Estimation Toolbox will be used for settings determination, due to its excellent performance in the preliminary investigations.

Validation, evaluation, and benchmarks

The success of the computational model will be evaluated against two factors; ability to replicate the in vitro system performance and the parameter estimation performance in elucidating settings for PhysioBank conditions simulation. The primary goal is to position this computational model as a bridge between the clinical conditions of interest to be simulated and the mock circulatory loop experimental design. Secondary is the models execution speed and the development of a multi-tiered approach to accelerating the execution of the primary goal. Accuracy: A Large Scale Trust-Region-Reflective Non Linear Least Squares method will be used as the assessment of waveform similarity during model fitting to patient data; this maintains consistency with the previous parameter estimation work [6668]. A normalized cost function based on sum of squared errors (SSE), will be used to assess the final similarity of the data: Equation 1

The cost function must be normalized to the number of discrete samples (n) used in the residuals (r) determination to account for the difference in beat periods for different heart rates. The threshold of 10 was developed during the preliminary work, and was found to produce acceptable results. Speed: The execution speed of the computational models should be greater than a real time simulation in the mock circulatory loop. A desired acceleration would be a 1:10, or larger, computational:experimental execution time ratio for a simulation. Timer methods within the Matlab environment will be used to assess the simulation performance.

Potential problems and alternate strategies

Full physical model execution speed cannot be accelerated. Investigation into model simplifications will be pursued; torque and pressure sources in place of mechanical and pneumatic systems [69]. Conversion to state-space models will be investigated if the simplified physical models are not sufficient in accelerating the computation time [70].

Parameter estimation converges on multiple solutions. In the event that the parameter estimation tools do not provide singular values for the settings associated with a particular waveform, reduction of the variables being estimated will become necessary. This will be achieved through time domain estimations of heart rate by periodicity calculations and percent systole through waveform inflection methods [71,72]. The remaining variables will be determined through the parameter estimation routine.

Aim 2: Deploy a fully automated mock circulatory loop

An in vitro testing platform is crucial for the investigation of devices destined for in vivo use. The testing system used for assessment of cardiac assist devices and prosthetic valves is the mock circulatory loop. The embodiment this design discussed in this proposal is a lumped parameter design targeted at simulating the



systemic circulation, specifically the conditions in the ascending aorta. The perspective of this work is the automation of the subsystems tasked with controlling each of these lumped parameters.

Experimental Approach

Producing the mock circulatory loop will be achieved through fabricating standalone subsystems and integrating them with a host controller. The subsystems deployment will involve design development, custom fabrication, and controller programming. A host controller will be developed to conduct the subsystems through the simulations of interest, and will act as a data repository during the acquisitions. Data handling solutions will be developed to ensure that experimental data is collected and stored efficiently [73].

The preliminary work on the peripheral resistor, arterial compliance chamber, and Harvard Apparatus pump illustrate the success of this investigator in producing the requisite testing subsystems. The design choices in these preliminary models took into account this final vision of integration, and were designed to be controlled by a host system. The distributed control approach was chosen for its robustness and the offloading computational power from a host system. Robustness of this design lies in the ability for a failed process control to have minimal impact on the rest of the system, with the host controller able to identify this failure and safely shutdown the system. The controller frequencies chosen to achieve the high performance of this system come at a computational cost, and exceed the capabilities of a Microsoft Windows based control system [74,75]. Distributing these high frequency tasks to embedded processors allows the host controller to have minimal computational overhead during experiments, which may be desirable if this mock loop controller is tasked with other running other investigational software (e.g. PIV programs).

1. Develop host controller that will coordinate the subsystems by communicating desired set points

selected by the user on a GUI interface or through the execution of a conditions file list. This host controller should also be able to integrate read back data from the subsystems with other sensor data being collected from the mock circulatory loop. The output of this host controller program should be in a format that is easily stored and manipulated. National Instruments LabVIEW programming environment will be utilized for its ease in creating GUIs (Figure C.9) and the well developed interfaces with data acquisition hardware currently owned by this laboratory. This environment also enables the host controller to be easily scaled to high performance Real-Time Operating Systems (RTOS) in the event a Windows based acquisition is not capable of handling the requirements of this application.

2. Integrate subsystems into a full systemic mock circulatory loop. This system embodiment will be comprised of the elements discussed in the preliminary data (peripheral resistor, arterial compliance chamber, automated Harvard Apparatus pump) and a venous reservoir (Figure C.10). The construction of an appropriate support scaffold and installing of conduit connections will be performed; a mechanical breadboard approach to the mounting and quick disconnect connections will be employed to ease future alterations. Ascending aortic condition simulation will be realized with this system version [76].

3. Fabricate the pulmonary simulator to appropriately construct the inlet conditions of the left ventricular simulator. This simulator will draw from the venous reservoir and deliver the appropriate pressure head, based

Figure C.9: Control dashboard for the mock circulatory loop. The fields in the upper left denote the conditions list file to run and the data output file. Sensor traces to the left are used for in process monitoring. The panels to the right interface to subsystems and show conditions progress.

Figure C.10: Mock circulatory loop currently deployed: Harvard Pump (1), aortic root model (2), compliance chamber (3), resistor (4), venous reservoir (5), pulmonary simulator (6), PIV camera (7), PIV laser rail (8).



on a flow delivered algorithm, replicating pulmonary output impedance. The hardware targeted for this subsystem is a pressure sensor downstream of a centrifugal pump and a level sensor in the venous reservoir feeding the pump; shown as element 6 in Figure C.10.

4. Install pressure and flow sensors between subsystems. Distributed pressure sensors and flow sensors along conduits in the system will be used for experimental performance evaluation: simulated arterial pressure, pulse wave velocity, measured cardiac output, and measured peripheral resistance. These sensors will be connected to the host controller via signal conditioning electronics and data acquisition hardware.

5. Implement a data storage solution that is capable of handling large acquisition times characteristic of ISO-14708 standards; 30 days in length [27]. The file structure envisioned is time oriented segmentation; a multi day investigation would result in sequential data files. Binary format and NI proprietary Technical Data Management Solution (TDMS) file format will be the initial capture file formats investigated. The data will have to be converted into the PhysioNet file format, with the appropriate channel names tagged, per the data sharing plan outlined in this proposal. This will allow for the PhysioBank ATM interface to be used for data visualization. Internal data storage redundancy will involve using a WebDAV server for online file storage with LZW compression of the files.


The validation of this in vitro system will come from its ability to produce physiologically accurate waveforms that were predicted by the computational model. Performance analysis will be conducted to ensure that the operating bandwidth of the system matches the specified design; limits tested with the isolated subsystems should be reflected in the complete mock circulatory loop. Reproducibility analysis will be performed to ensure that the system can accurately maintain set points to produce highly similar pressure waveforms. Accuracy: The cost function analysis (Equation 1) used in the computational to experimental analysis (Aim 1) will be used for reproducibility verification; the threshold of



fabrication method and material choice. The preliminary work substantiated swept CAD model design based on traced cryo-slice data, and machining the lumen from acrylic to achieve the aforementioned design requirements. Applying this method to the left atrium is seen as a viable means of delivering a model of this anatomy. This subtractive machining yields a smooth surface with a clear wall that requires minimal polishing prior to imaging, as opposed to SLA models with their crosshatched wall construction. The structural integrity of this machined model is also superior to deposition models, which is critical when considering the handling to gain access to the prosthetic valves being tested [79,80].

1. Construct left atrium computer model from the Visible Human Project female cryo slice data. ITK-SNAP will be used to trace the boundary of the left atrium from the mitral valve plane to the ostias of the various interfacing vessels. Geometry past the ostia will not be considered, as this extends the model beyond what is considered to be sufficient at this stage. Blender will be used to refine the results of the ITK-SNAP tracing; smoothing of the mesh and facet reduction will be performed. The processed mesh will be imported to SolidWorks for development of the swept model. This swept model allows for smooth surface machining and reduces the computational effort of working with this geometry in the CAD packages. Interfaces will be designed to appropriately connect the fabricated model to the mock circulatory loop and allow for different mitral valves to be inserted for analysis.

2. Fabrication of the left atrium model from cast acrylic. The swept model will be processed into a split part design to allow for the machining of separate interfacing pieces. MasterCAM for SolidWorks will be used to program the machine paths. The same machining technique from the preliminary work on the ascending aorta will be applied to this model; rough cut material removal, waterline finishing passes, and scallop finishing passes. The model pieces will be adhered at the part lines to construct the full left atrium. Subsequent machining of the interfacial geometry will be performed to allow for mock loop connection and prosthetic mitral valve placement.

3. Incorporate left atrium model in mock circulatory loop. Connection of the model to the mock circulatory loop will be achieved through the designed connections for the mitral valve interface and the pulmonary veins. Scaffolding will have to be engineered to adequately support the model on the deck, which should also allow for the PIV system to gain access to the investigation space.


Successful fabrication of the left atrium model, with subsequent imaging of left atrial flow patterns by PIV, will validate the completion of this proposed aim. Assessment of this model will be taken from a functional to performance perspective. Adequate hermetic seals will be validated through hydraulic pressure tests when connected to the mock circulatory loops; this will test not only the connection seals but the part line seal. It is critical that this model holds its hermetic seals for a range of -150 to 250 mmHg gauge pressure; this is the range of pressures that could exist upstream of the Harvard Apparatus pump [81]. Accuracy: Model will be verified to have similar boundaries to that of the lumen in the cryo slice data. Post machining CMM will be used to verify the machining paths, and could be used to validate against the original cryo slice geometry for full workflow verification. Performance: Acquisition of images that clearly depict particle movement in the flow will validate the PIV compatibility of the model. Assessment will be performed with a 40:60 (weight) glycerine:water solution, which will result in slight distortion due to the refractive index mismatch [82]. Only clarity of the model is being assessed, as refractive index matching is not needed to ensure PIV compatibility.

Potential problems and alternate strategies

Left atrium ostias cannot be machined with a single part line due to the limitations of the CNC method. Part line choice will be investigated, as another plane of dissection could provide the appropriate access by the mill. Multiple part lines will also be considered. Subsequent outsourcing of the work to a shop with 5-axis mill capabilities will result if the in-house equipment limitations cannot be overcome.

Left atrium cannot be machined by a split part design. The model will be fabricated in ABS plastic using the in house deposition printing equipment. PIV windows machined from acrylic will be inserted into the printed model to satisfy the imaging requirements. Utilization of SLA technology could be applied if the window method is deemed insufficient and research funds support this decision. Limitations in structural integrity will have to be entertained if deposition models are used, with possible engineering solutions needed to prevent material failure due to use.


Proprietary and Confidential Bibliography and References Cited Page 21 of 25

Bibliography and References Cited

[1] Neal M. L., and Kerckhoffs R., 2010, Current progress in patient-specific modeling, Brief Bioinform, 11(1), pp. 111126.

[2] Patient Simulator and Training Center (Mock Circulation Tank); Users Manual [Online]. Available: http://www.perfusion.ws/Syncardia/TAH-t%20Training%20Center%20User%20Manual.pdf. [Accessed: 16-Dec-2012].

[3] Iaizzo P. A., 2009, Handbook of cardiac anatomy, physiology, and devices, Springer, New York, NY. [4] Yubing S., Patricia L., and Rodney H., Review of Zero-D and 1-D Models of Blood Flow in the

Cardiovascular System, BioMedical Engineering OnLine, 10. [5] Kar B., Delgado III R. M., Frazier O., Gregoric I. D., Harting M. T., Wadia Y., Myers T. J., Moser R. D.,

and Freund J., 2005, The effect of LVAD aortic outflow-graft placement on hemodynamics and flow: Implantation technique and computer flow modeling, Texas Heart Institute Journal, 32(3), p. 294.

[6] Long Q., Merrifield R., Yang G. Z., Kilner P. J., Firmin D. N., and Xu X. Y., 2003, The influence of inflow boundary conditions on intra left ventricle flow predictions., Journal of biomechanical engineering, 125(6), p. 922.

[7] Patel S. M., Allaire P. E., Wood H. G., Throckmorton A. L., Tribble C. G., and Olsen D. B., 2005, Methods of failure and reliability assessment for mechanical heart pumps, Artificial organs, 29(1), pp. 1525.

[8] Moody G. B., Mark R. G., and Goldberger A. L., 2011, PhysioNet: Physiologic signals, time series and related open source software for basic, clinical, and applied research, Proceedings of the 33rd Annual International Conference of the IEEE EMBS, Boston, Massachusetts, pp. 83278330.

[9] Ackerman M. J., 1998, The visible human project, Proceedings of the IEEE, 86(3), pp. 504511. [10] Liu Y., 2003, Design Of A Performance Test Loop For An Artificial Heart Pump, Masters Thesis,

University of Virginia, Charlottesvill, VA USA. [11] Timms D. L., Gregory S. D., Greatrex N. A., Pearcy M. J., Fraser J. F., and Steinseifer U., 2011, A

Compact Mock Circulation Loop for the In Vitro Testing of Cardiovascular Devices, Artificial Organs, 35(4), pp. 384391.

[12] Hsu P. L., Cheng S. J., Saumarez R. C., Dawes W. N., and McMahon R. A., 2008, An extended computational model of the circulatory system for designing ventricular assist devices, ASAIO Journal, 54(6), p. 594.

[13] Lvine J., and Mllhaupt P., eds., 2010, Advances in the theory of control, signals and systems with physical modeling, Springer, Berlin.

[14] Johansson P., and Andersson B., 2008, Comparison of simulation programs for supercapacitor modelling, Masters Thesis, Chalmers University of Technology, Gothenberg, Sweden.

[15] Gregory S. D., 2009, Simulation and development of a mock circulation loop with variable compliance, Masters Thesis, Queensland University of Technology, Brisbane, Australia.

[16] Pantalos G. M., Koenig S. C., Gillars K. J., Giridharan G. A., and Ewert D. L., 2004, Characterization of an Adult Mock Circulation for Testing Cardiac Support Devices, ASAIO Journal, 50(1), pp. 3746.

[17] De Zlicourt D., Pekkan K., Kitajima H., Frakes D., and Yoganathan A. P., 2005, Single-Step Stereolithography of Complex Anatomical Models for Optical Flow Measurements, Journal of Biomechanical Engineering, 127(1), p. 204.

[18] Kornecki A., and Zalewski J., 2009, Certification of software for real-time safety-critical systems: state of the art, Innovations in Systems and Software Engineering, 5(2), pp. 149161.

[19] Severa O., and Cech M., 2012, REX - Rapid development tool for automation and robotics, 2012 IEEE/ASME International Conference on Mechatronics and Embedded Systems and Applications (MESA), pp. 184 189.

[20] Loire J., and Neelakantam A. K. R., 2011, Development of a Steer-By-Wire application in MATLAB/Simulink, Masters Thesis, Chalmers University of Technology, Gothenberg, Sweden.

[21] Li J. K.-J., and Zhu Y., 1994, Arterial Compliance and Its Pressure Dependence in Hypertension and Vasodilation, Angiology, 45(2), pp. 113117.

[22] Panwar N. S., Kelly G. S., Taylor C. E., and Miller G. E., 2012, Search Engine of Physiological Conditions in the PhysioBank Database., Podium presentation at the Virginia Academy of Science 90th Annual Meeting, Norfolk, VA.

[23] Manning K. B., and Miller G. E., 1999, Shaft/Shaft-Seal Interface Characteristics of a Multiple Disk Centrifugal Blood Pump, Artificial Organs, 23(6), pp. 552558.


Proprietary and Confidential Bibliography and References Cited Page 22 of 25

[24] Thatte S. M., 2006, In Vitro Flow Visualization Study of the Interface Between Outflow Graft of Ventricular Assist Device and Aorta, Dissertation, Virginia Commonwealth University, Richmond, VA USA.

[25] Walker A., 2010, In Vitro Evaluation of Mechanical Heart Valve Performance Using a Novel Test Chamber in an Automated Mock Circulatory Loop, Masters Thesis, Virginia Commonwealth University, Richmond, VA USA.

[26] International Organization for Standardization, 1987, Centrifugal, mixed flow and axial pumps - Code for hydraulic performance tests - Precision class, ISO 5198.

[27] International Organization for Standardization, 2010, Implants for surgery - Active implantable medical devices, Part 5: Circulatory support Devices, ISO 14708-5.

[28] International Organization for Standardization, 2005, Cardiovascular implants Cardiac valve prostheses, ISO 5840.

[29] Taylor C. E., Rourick R. A., and Walsh J. P., 2007, High Throughput Solubility Analysis; a Tailored Approach to Supporting Drug Discovery Efforts, Poster presentation at the American Association of Pharmaceutical Scientists (AAPS) Annual Meeting and Exposition, San Diego, CA.

[30] Taylor C. E., Rourick R. A., and Walsh J. P., 2007, A Modular Chemical and Photo-Stability Chamber Purpose-Built for an LC/MS-Enabled Liquid Handler, Poster presentation at the American Association of Pharmaceutical Scientists (AAPS) Annual Meeting and Exposition, San Diego, CA.

[31] Taylor C. E., Yang Q., Walsh J. P., and Rourick R. A., 2007, An Automated Dynamic BioAnalytical Sample Preparation Routine for Integration with Liquid Handlers, Poster presentation at the Chemical and Pharmaceutical Structure Analysis (CPSA) Conference, Langhorne, PA.

[32] Taylor C. E., and Miller G. E., 2012, Implementation of an automated peripheral resistance device in a mock circulatory loop with characterization of performance values using Simulink Simscape and Parameter Estimation., Journal of Medical Devices, 6(4), p. 045001.

[33] Wilmer W. Nichols, 2011, McDonalds blood flow in arteries: theoretical, experimental and clinical principles., Hodder Arnold, London.

[34] Watton J., 2009, Fundamentals of fluid power control, Cambridge University Press, Cambridge UK;;New York.

[35] Liptk B., 2006, Instrument engineers handbook, CRC Taylor & Francis, Boca Raton Fla. [u.a.]. [36] Taylor C. E., Dziczkowski Z. W., and Miller G. E., 2012, Automation of the Harvard Apparatus Pulsatile

Blood Pump, Journal of Medical Devices, 6(4), p. 045002. [37] Jamali P., and Shirazi K. H., 2012, Robot Manipulators: Modeling, Simulation and Optimal Multi-Variable

Control, Applied Mechanics and Materials, 232, pp. 383387. [38] Zarghani N. H., 2009, Modeling and simulation of an active robotic device for flexible needle insertion,

Master, National University of Singapore, Singapore. [39] Liu Y., Allaire P., Wood H., and Olsen D., 2005, Design and initial testing of a mock human circulatory

loop for left ventricular assist device performance testing, Artificial Organs, 29(4), pp. 341345. [40] Liu Y., Allaire P., Wu Y., Wood H., and Olsen D., 2006, Construction of an Artificial Heart Pump

Performance Test System, Cardiovascular Engineering, 6(4), pp. 151158. [41] Taylor C. E., and Miller G. E., 2012, Mock Circulatory Loop Compliance Chamber Employing a Novel

Real-Time Control Process, Journal of Medical Devices, 6(4), p. 045003. [42] Ugural A., 1999, Stresses in plates and shells, WCB/McGraw Hill, Boston. [43] Reddy J., 1999, Theory and analysis of elastic plates, Taylor & Francis, Philadelphia PA. [44] Reddy J. N., Huang C. L., and Singh I. R., 1981, Large deflections and large amplitude vibrations of

axisymmetric circular plates, Int. J. Numer.

Documents

Dissertation Proposal