Student Projects Archive

2022 2021 2019  2018  2017  Current Projects

2022

Modeling Arteriovenous Fistula Hemodynamics in ESRD Patients with Pulmonary Hypertension

Faculty Mentor:  Fatemeh Bahmani, PhD, Postdoctoral Fellow, Department of Engineering

Permanent structural damage to the kidneys results in chronic kidney disease, a condition affecting over 15% of Americans. Over time, this can develop into end-stage renal disease (ESRD) where patients require hemodialysis, through a surgically-created arteriovenous fistula, to filter and purify circulating blood. One comorbidity of ESRD is pre-capillary pulmonary hypertension (PH) which is believed to be connected to the creation of the arteriovenous fistulas; however, there is little conclusive evidence supporting this hypothesis. The overall objective of this project is to correlate fistula and pulmonary artery hemodynamics to better understand PH development in ESRD patients and engage in a preliminary analyses to connect these parameters to standard clinical measures for early identification and monitoring.  Previous REU projects have focused on modeling hemodynamics in the pulmonary artery. Here, we focus on modeling fistula flow in patients with pulmonary hypertension. To accomplish these purposes, the student will: 1) use Mimics (Materialise) to segment fistula geometry from MR images; 2) mesh geometries using ANSYS Workbench; 3) develop computational fluid dynamics models (ANSYS FLUENT) with appropriate boundary conditions; 3) visualize results; and 4) compare fistula flow parameters to PH metrics (pulmonary artery pressure and vascular resistance).

Effect of concurrent visual and auditory attention tasks on brain activity during postural control

Faculty Mentor:  James Lin, PhD, PT, MS, Assistant Professor, Department of Physical Therapy

Project description: Cognitive demands for human postural control, especially attention, can be evaluated by dual-task paradigms, in which a secondary cognitive task is performed spontaneously with a balance task. Previous studies have used auditory tasks as the secondary cognitive task to examine the effect of attention on human postural control. However, besides the auditory task, people are occupied by various visual attention tasks in daily life. Little is known regarding how the human brain processes concurrent auditory and visual attention tasks while maintaining postural control. The purpose of Dr. Lin’s project is to investigate brain activity during concurrent visual and auditory attention tasks in a standing position. Functional near-infrared spectroscopy (fNIRS) will be used to detect brain activity and visual attention tasks will be provided via a virtual reality headset. Specifically, students will be expected to: 1) complete research training for the human subjects research and project specific tasks; 2) help with data collection and processing; and 3) conduct analyses to determine the effect of concurrent visual and attention on human postural control.

Uncovering Mechanisms of Synapse Formation in Human Brain Models

Faculty Mentor: Karen Litwa, PhD, Assistant Professor, Department of Anatomy and Cell Biology, Brody School of Medicine

Synapses are the sites of information transfer between neurons and are responsible for the emergence of complex cognitive functions. We research the molecular mechanisms of synapse formation in the prenatal human brain. Both genetic mutations and environmental factors that disrupt synapse formation result in neurodevelopmental disorders. We are particularly interested in excitatory synapses, which facilitate information transfer through action potentials. Excitatory synapses form between pre-synaptic axons and specialized mushroom-shaped post-synaptic projections, known as dendritic spines. However, most spines form postnatally.  It is still unclear how synapses initially form. We hypothesize that filopodia-like projections emanating from dendrites serve as the initial sites of synapse formation.  Our research is aimed at elucidating the molecular requirements for dendritic filopodia to initiate synapse formation and subsequently mature in mushroom-shaped spines. To conduct this research, students will use a combination of cutting-edge technologies, including 1) tissue engineering to generate human cortical brain spheroids, 2) fluorescence microscopy techniques including super-resolution STORM imaging and analysis, and 3) CRISPR transcriptional manipulation.

Characterization of Drosophila Heartbeat Parameters Using a High-Speed Microscopic Imaging System

Faculty Mentor: Yang Liu, PhD, Assistant Professor, Department of Engineering 

Western diet pattern has been considered as the main source of many heart diseases. However, it is a long-term effect that can not be tested and demonstrated on human bodies under controlled conditions. Drosophila has been proven to be an effective genetic model system that can be used to gain information about genetic regulation for the early developmental stages of the hearts of vertebrates. Several dietary patterns that affect the heart function of Drosophila may have close parallels to human heart diseases. By examining the effects of different dietary patterns on the heart function of Drosophila, we will be able to advance the knowledge about the effects of those diet patterns on human heart functions. In this project, we will characterize the Drosophila heartbeat parameters by using a high-speed microscopic imaging system. The recorded heartbeat images will be imported into a custom MATLAB code for further analysis. The attributes extracted from the images will include M-mode of heartbeat, separation distance and velocity of heart tube inner walls, and heart rate.

Development and Validation of a Simplified Approach to Modeling Bone Stress

Faculty Mentor: Stacey A. Meardon, PhD, PT, Associate Professor, Department of Physical Therapy

Metatarsal and tibia stress injuries are common in tactical, recreational and competitive athletes. The ability to estimate subject specific metatarsal and tibial stresses will provide a basis for understanding loads experienced during physical activity, potentially influencing rehabilitative and preventative efforts.  Metatarsal and tibia loads experienced by the human body during physical activity can be obtained through a combination of experimental data collection and musculoskeletal modeling. Ongoing work aims to identify efficient and valid approaches to estimate bone stress in large scale studies of physically active populations. Subject-specific cross-sectional bone stress during lower limb motor tasks, estimated using a series of musculoskeletal models, will be cross validated with cadaver strain gage data, full tibial finite element models, and/or existing literature values. For this project, the student will: 1) assist in collecting and processing 3D motion data; 2) process MRI, CT images, and/or US data using processing tools in MATLAB; 3) develop 3D cross-sectional metatarsal and tibia finite element bone meshes using single slice imaging at regions of interest; and 4) compare stress to full finite element model data.

Modeling of Cortical Dynamical Networks in Motor Control

Faculty Mentor: Chris Mizelle, PhD, Associate Professor, Department of Kinesiology

Right-handed individuals activate networks in the left side of the brain while performing motor acts with their right hand. It has been assumed that left-handed individuals would show brain activations that “mirror” this; However, new research has raised doubts. To understand the motor neurophysiology in left-handed individuals, and how these mechanisms differ from right-handed individuals, direct study of neural activations is needed. Traditional measures of brain activations, acquired using electroencephalography (EEG), focus on evaluation of time-voltage and time-frequency analyses of sensor-level data. Recent computational developments, however, provide researchers with advanced techniques to study brain activation localizations and information flow dynamics. Past REU students demonstrated feasibility. Extending previous work, this project will use EEG to image neural activations and develop a neural networks model to describe the information flow in left- and right-handed individuals. To accomplish this purpose, the student will: 1) collect EEG data; 2) implement information flow measures in MATLAB; 3) validate the neural network model by evaluating information flow measures in different behavioral domains; and 4) compare information flow dynamics and source localization results between left- and right-handed individuals.

Models of Visual Scanning for Concussion Testing

Faculty Mentor: Nicholas Murray, PhD, Professor, Department of Kinesiology

The oculomotor system can be an indicator of the neurological status of an individual.  For example, individuals with traumatic brain injury (TBI) and mild traumatic brain injury (mTBI) suffer from a number of vision and visual processing dysfunctions, including visual field defect, vision motion sensitivity, and oculomotor deficits. Saccadic eye movement assessment provides a biological marker for specific pathology and indicates cortical as well as subcortical brain areas that influence motor performance: It is proposed as an endophenotypic marker of the injury state.  While there is considerable evidence for motor dysfunction following brain injury, the purpose of this project is to model visual scan patterns of individuals with and without head injury using an eye tracking enabled Virtual Reality (VR) system.  Specifically, the student will: 1) use MATLAB or similar program to develop analytical measures for brain injury within an eye tracking enabled VR system; 2) analyze visual control parameters that are most predictive of brain injury; and 3) develop models using visual control variables to distinguish long term effects of head injury.

Computational analysis of brain activation: an EEG study on vocal disgust and facial attractiveness

Faculty Mentor: Kathrin Rothermich, PhD, Assistant Professor, Department of Communication Sciences and Disorders

When analyzing functional neuroimaging data, especially electroencephalography (EEG), one challenge is the temporal and spatial overlap of event-related potentials (ERPs). Principal component analysis (PCA) is a useful computational tool for the statistical decomposition of ERPs and helps to solve this issue. Using PCA, the current project focuses on the cross-modal priming effect of vocal disgust on the processing of facial attractiveness. When making evaluative judgements about other people, we are often influenced by the surrounding context, such as smells or images. Based on this context, we might be biased to give either positive or negative evaluations, for example when judging physical attractiveness. One emotion that has been shown to influence attractiveness judgements is disgust. While different stimuli, such as images or smells, have been used to study the effects of disgust, fewer studies have investigated the effect of acoustic stimuli, such as vocalizations (for example “ugh” or “ eww ”). Thus, it is unclear whether disgust vocalizations can influence attractiveness judgments. While the EEG data have been collected, we have yet to analyze the EEG data and compute the statistical comparisons. To accomplish the study purpose, the student will: 1) implement the preprocessing of electrophysiological data using EEGLab in MATLAB; 2) learn how to use computational tools, such as PCA, to decompose the ERPs in EEGLab into meaningful outcomes; and 3) become proficient at analyzing ERPs using ERPLab.

Neural Basis of Design Fixation

Faculty Mentor: Brian Sylcott, PhD, Associate Professor, Department of Engineering

Design fixation refers to blind adherence to a set of ideas, which can limit the output of conceptual design. Engineering designers tend to fixate on features of pre-existing solutions and consequently generate designs with similar features. The lack of variety resulting from this behavior can inhibit design innovation. As such, researchers have sought to better understand the mechanisms behind design fixation in order to develop strategies to mitigate its effects. While traditional attempts to study design fixation have focused on behavioral observations, advances in neuroimaging have opened new avenues for research. This project will employ a novel neuroimaging technology, functional near-infrared spectroscopy (fNIRS), to study the brain activity of engineering designers during conceptual design in order to understand how design fixation is reflected in a person’s brain when solving design problems. Preliminary functional magnetic resonance imaging (fMRI) results show increased activation in areas associated with visuospatial processing when comparing ideation activities completed using a fixation source to those without one. Activation was found in the right inferior temporal gyrus, left middle occipital gyrus, and right superior parietal lobule regions. The left lingual and superior frontal gyri were found to be less active in the example condition; these gyri are close in proximity to the prefrontal cortex, associated with creative output. The spatial patterns of activation provide evidence that a shift in mental resources can occur when a designer becomes fixated. These results will be explored further in this work. Specifically, students will: 1) collect neurophysiological data from healthy adults completing conceptual design tasks; 2) process fNIRS and behavioral data; and 3) correlate data with experimental conditions.

Stability Fluctuations during Locomotion

Faculty Mentor: Ryan Wedge, PT, PhD, Assistant Professor, Department of Physical Therapy

People may optimize gait stability when they walk with preferred patterns. It is unclear how gait stability is modulated when doing different gait tasks, such as traversing uneven terrain. Also, it is unknown if people can further lower the amount of necessary stability when walking with preferred speed on unthreatening terrain. The end goal of this project is to develop interventions that may improve stability in people with pathology, such as lower limb amputation or stroke. The aim of the current step in this project is to develop a real-time stability feedback system that can be used in the lab environment. To accomplish this aim and in collaboration with the research team, the student will: 1) Collect and analyze motion capture data necessary for the stability calculation using Qualisys, V3D, MATLAB and LabVIEW, 2) Create a program for multiple stability metrics in real-time, and 3) test the program with unthreatened and threatened walking.

2021

Modeling of Activation Specific Muscle Moment Arms

Faculty Mentor: Zachary Domire, PhD, Associate Professor, Department of Kinesiology

Muscle moment arms are a critical component of musculoskeletal models, not only for calculating torque, but also for determining muscle fiber excursion. Past work from our previous REU grants has shown that moment arms are a critical component to improving the accuracy of subject specific models. Other past REU projects have refined speckle tracking algorithms to improve calculation of moment arms from ultrasound data. There is evidence in the literature that moment arms can vary based on muscle activation. The purpose of this project is to measure moment arms at varying muscle activation and model this relationship. Muscle models will be developed with these data and the models will be validated by comparing output to experimental measured strength curves. For this project, the student will: 1) collect ultrasound data from the Aixplorer ultrasound system on the lower extremity musculature; 2) refine MATLAB routines for calculation of muscle moment arms and fascicle lengths; 3) measure isometric strength throughout the range of motion, while measuring muscle activation using EMG; 4) input parameters into a muscle model developed in MATLAB; and 5) compare model results with experimentally measured strength.

Development and Validation of a Simplified Approach to Modeling Bone Stress

Faculty Mentor: Stacey A. Meardon, Assistant Professor, Department of Physical Therapy

Metatarsal and tibia stress injuries are common in tactical, recreational and competitive athletes. The ability to estimate subject specific metatarsal and tibial stresses will provide a basis for understanding loads experienced during physical activity, potentially influencing rehabilitative and preventative efforts.  Metatarsal and tibia loads experienced by the human body during physical activity can be obtained through a combination of experimental data collection and musculoskeletal modeling. Ongoing work aims to identify efficient and valid approaches to estimate bone stress in large scale studies of physically active populations. Subject-specific cross-sectional bone stress during lower limb motor tasks, estimated using a series of musculoskeletal models, will be cross validated with cadaver strain gage data, full tibial finite element models, and/or existing literature values. For this project, the student will: 1) assist in collecting and processing 3D motion data; 2) process MRI, CT images, and/or US data using processing tools in MATLAB; 3) develop 3D cross-sectional metatarsal and tibia finite element bone meshes using single slice imaging at regions of interest; and 4) compare stress to full finite element model data.

Fibroblast-mediated myocardial scar maturation and stability

Faculty Mentor: Lisandra de Castro Brás, PhD Department of Physiology

Maladaptive remodeling of the left ventricle (LV) during disease, or after injury such as myocardial infarction (MI, also known as heart attack), is often a cause of heart faillure. Maladaptive LV remodeling leads to the formation of a collagenous scar, whose makeup (size, composition, vascularization, and biomechanical properties) is a critical determinant of patient outcomes. While the importance of scar quality has long been recognized, current therapies are targeted mostly at reducing scar size – either by increasing myocyte survival or by mitigating inflammation and fibrosis. Paradoxically, suppression of inflammation and fibrosis has led to unexpected results, including infarct expansion and LV wall thinning; thus, simply reducing scar size seems to be inadequate.  We propose to shift the current paradigm of just focusing on scar size reduction post injury, by including new focus on scar quality (i.e. composition, vascularization, and biomechanical properties) to prevent adverse LV remodeling through scar stabilization.

For this purpose, we will use a peptide derived from cardiac extracellular matrix named p1158/59. Our preliminary data support a role for p1158/59 in fibroblast function – the cells responsible for scar formation. Activation of fibroblasts, into a myofibroblast phenotype, is necessary for tissue healing; however, excessive or uncontrolled activation leads to tissue fibrosis and cardiac dysfuntion. Our central hypothesis is that p1158/59-fibroblast signaling promotes scar stabilization after cardiac injury by reducing myofibroblast activation; this in turn leads to changes in the composition, vascularization, and biomechanical properties of the scar. Using an MI model, we will test our hypothesis first by assessing whether p1158/59 therapy alters myocardial scar content, vascularization, and/or biomechanical properties.

Investigating the Role of Inherent Fiber Tension in the Fibrinolytic Susceptibility of Blood Clot Constituents

Faculty Mentor: Nathan Hudson, PhD, Assistant Professor, Dept. of Physics

Proper wound healing necessitates both coagulation (the formation of a blood clot) and fibrinolysis (the dissolution of a blood clot). A thrombus resistant to clot dissolution can obstruct blood flow, leading to vascular pathologies. We recently showed that fibrin fibers, the main structural constituents of blood clots, polymerize in a tensed state.  This study seeks to understand the role played by this inherent fibrin fiber tension in the dissolution of a blood clot. We will use fluoresce microscopy and potentially microfluidics-based approaches to study this system.

Modeling the Functional Consequences of Asymmetrical Strength

Faculty Mentor: Anthony S. Kulas, PhD, Associate Professor, Dept. of Health Education and Promotion

Persistent quadriceps and hamstring muscle strength deficits are common following ACL reconstruction and are dependent on the site of graft harvest.  The prevalent consensus is that a ≤10% strength deficit is important for a successful post-surgical outcome. However, the magnitude of strength deficits that manifest as functional deficits, thus placing an individual at risk for a second ACL injury, are unclear.  Previous work has shown that reduced semitendinosus tendon stiffness, a common harvest tissue, effectively reduced hamstring muscle moments during simulated running.  We aim to extend this work, by modeling the functional effects of whole muscle group strength deficits. The purpose of this project is to determine the effect of strength deficit on lower extremity muscle moments during single leg hopping for distance, a common functional task evaluated following ACL reconstruction. For this project, the student will: 1) simulate single leg hopping for distance in OpenSim; 2) model muscle strength deficits in increments of 5%, 10%, 15%, and 20%; and 3) determine the effect of these modeled strength deficits on hip, knee, and ankle joint moments.

Modeling of Cortical Dynamical Networks in Motor Control

Faculty Mentor: Chris Mizelle, PhD, Assistant Professor, Department of Kinesiology

Right-handed individuals activate networks in the left side of the brain while performing motor acts with their right hand. It has been assumed that left-handed individuals would show brain activations that “mirror” this; However, new research has raised doubts. To understand the motor neurophysiology in left-handed individuals, and how these mechanisms differ from right-handed individuals, direct study of neural activations is needed. Traditional measures of brain activations, acquired using electroencephalography (EEG), focus on evaluation of time-voltage and time-frequency analyses of sensor-level data. Recent computational developments, however, provide researchers with advanced techniques to study brain activation localizations and information flow dynamics. Past REU students demonstrated feasibility. Extending previous work, this project will use EEG to image neural activations and develop a neural networks model to describe the information flow in left- and right-handed individuals. To accomplish this purpose, the student will: 1) collect EEG data; 2) implement information flow measures in MATLAB; 3) validate the neural network model by evaluating information flow measures in different behavioral domains; and 4) compare information flow dynamics and source localization results between left- and right-handed individuals.

Models of Visual Scanning for Concussion Testing

Faculty Mentor:  Nicholas Murray, PhD, Professor, Department of Kinesiology

The oculomotor system can be an indicator of the neurological status of an individual.  For example, individuals with traumatic brain injury (TBI) and mild traumatic brain injury (mTBI) suffer from a number of vision and visual processing dysfunctions, including visual field defect, vision motion sensitivity, and oculomotor deficits. Saccadic eye movement assessment provides a biological marker for specific pathology and indicates cortical as well as subcortical brain areas that influence motor performance: It is proposed as an endophenotypic marker of the injury state.  While there is considerable evidence for motor dysfunction following brain injury, the purpose of this project is to model visual scan patterns of individuals with and without head injury using an eye tracking enabled Virtual Reality (VR) system.  Specifically, the student will: 1) use MATLAB or similar program to develop analytical measures for brain injury within an eye tracking enabled VR system; 2) analyze visual control parameters that are most predictive of brain injury; and 3) develop models using visual control variables to distinguish long term effects of head injury.

Utilizing fNIRS to Model Sensory Integration During Physical and Cognitive Tasks

Faculty Mentor: Brian Sylcott, PhD, Assistant Professor, Department of Engineering

In humans, the vestibular system is a set of sensors that allows us to maintain a sense of balance. The underlying mechanisms behind how the human brain interprets sensory information during the sensory integration process are not well-understood. Furthermore, there is limited data on performing cognitive tasks while concurrently interpreting large amounts of sensory data during constant vestibular stimulation. To address this gap in knowledge, this research will employ a novel neuroimaging technology, functional near-infrared spectroscopy (fNIRS), and eye-tracking technology, to characterize attention during cognitive tasks. It was hypothesized that processing a large amount of sensory information from optical flow will impede a subject’s ability to perform tasks. Although VR generated optical flow was validated, preliminary results from REU work found increased prefrontal cortex activity and no significant difference in cognitive performance relative to optical flow speed. These unexpected results will be explored further in subsequent work leading to a mechanistic model of sensory integration. Specifically, students will: 1) collect neurophysiological data from healthy adults completing physical and cognitive tasks; 2) process fNIRS and eye-tracking data; and 3) correlate data with optical flow speeds.

Effect of concurrent visual and auditory attention tasks on brain activity during postural control

Faculty Mentor: James Lin, PhD, PT, MS, Assistant Professor, Department of Physical Therapy

Project description: Cognitive demands for human postural control, especially attention, can be evaluated by dual-task paradigms, in which a secondary cognitive task is performed spontaneously with a balance task. Previous studies have used auditory tasks as the secondary cognitive task to examine the effect of attention on human postural control. However, besides the auditory task, people are occupied by various visual attention tasks in daily life. Little is known regarding how the human brain processes concurrent auditory and visual attention tasks while maintaining postural control. The purpose of Dr. Lin’s project is to investigate brain activity during concurrent visual and auditory attention tasks in a standing position. Functional near-infrared spectroscopy (fNIRS) will be used to detect brain activity and visual attention tasks will be provided via a virtual reality headset. Specifically, students will be expected to: 1) complete research training for the human subjects research and project specific tasks; 2) help with data collection and processing; and 3) conduct analyses to determine the effect of concurrent visual and attention on human postural control.

Energy Expenditure Optimization during Locomotion

Faculty Mentor: Ryan Wedge, PT, PhD, Assistant Professor, Department of Physical Therapy

People optimize metabolic energy expenditure when they walk with preferred patterns. It is unclear how energy expenditure is modulated when doing different gait tasks, such as traversing uneven terrain. Also, it is unknown if people can further lower metabolic energy expenditure when walking with preferred speed on unthreatening terrain. The end goal of this project is to develop interventions that may improve energy expenditure in people with pathology, such as lower limb amputation or stroke. The aim of the current step in this project is to develop a real-time metabolic energy expenditure feedback system that can be used in the lab environment. The real time feedback system will be compared to indirect calorimetry, which is commonly used in biomechanics laboratories but cannot provide real time feedback of energy being expended. To accomplish this aim and in collaboration with the research team, the student will: 1) Collect and analyze muscle activity and motion capture data necessary for the energy expenditure surrogate using Qualisys, V3D, MATLAB and LabVIEW, 2) Create new and better metabolic surrogates, and 3) Analyze the correlation of metabolic surrogates to indirect calorimetry.

Computational analysis of brain activation: an EEG study on vocal disgust and facial attractiveness

Faculty Mentor: Kathrin Rothermich, PhD, Assistant Professor, Department of Communication Sciences and Disorders

When analyzing functional neuroimaging data, especially electroencephalography (EEG), one challenge is the temporal and spatial overlap of event-related potentials (ERPs). Principal component analysis (PCA) is a useful computational tool for the statistical decomposition of ERPs and helps to solve this issue. Using PCA, the current project focuses on the cross-modal priming effect of vocal disgust on the processing of facial attractiveness. When making evaluative judgements about other people, we are often influenced by the surrounding context, such as smells or images. Based on this context, we might be biased to give either positive or negative evaluations, for example when judging physical attractiveness. One emotion that has been shown to influence attractiveness judgements is disgust. While different stimuli, such as images or smells, have been used to study the effects of disgust, fewer studies have investigated the effect of acoustic stimuli, such as vocalizations (for example “ugh” or “ eww ”). Thus, it is unclear whether disgust vocalizations can influence attractiveness judgments. While the EEG data have been collected, we have yet to analyze the EEG data and compute the statistical comparisons. To accomplish the study purpose, the student will: 1) implement the preprocessing of electrophysiological data using EEGLab in MATLAB; 2) learn how to use computational tools, such as PCA, to decompose the ERPs in EEGLab into meaningful outcomes; and 3) become proficient at analyzing ERPs using ERPLab.

2019

Utilizing a physical Windkessel model in a coronary artery bypass graft phantom model with comparison to computational fluid dynamics

Faculty Mentor: Dr. Stephanie George, Assistant Professor Department of Engineering

Computational Fluid Dynamics (CFD) models, combined with imaging, allow for the development of subject, or phantom specific hemodynamic models. These models may be used to diagnose, understand underlying mechanisms of disease, evaluate treatment success, evaluate and test surgical procedures, or predict clinical events or outcomes. Previous REU students have developed CFD models in the pulmonary artery and idealized coronary bypass graft (CABG) models. Many CFD studies have examined the geometry of the anastomoses and its effect on competitive flow, however few have considered the downstream effects of perfusion/resistance and collateral flow.  A previous REU student investigated the impact of collateral flow on CABG.  The goal of this REU project is to incorporate a physical Windkessel model (resistance component) into a CABG flow loop and compare flow measured with a novel optical imaging method, iCertainty™, and CFD model. To accomplish these purposes, the student will: 1) build a physical Windkessel model; 2) collect experimental flow data using iCertainty™, 3) create a solid model of the phantom and mesh the fluid volume using ANSYS Workbench; 3) develop the CFD cases (ANSYS FLUENT) with appropriate boundary conditions from iCertainty™; 4) visualize results using ANSYS Post-Processing; and 5) compare flow from iCertainty™ and CFD results.

Subject specific modeling of the plantar flexors to enable study of Achilles Tendon rupture rehabilitation.

Faculty Mentor: Dr. Zachary Domire, Associate Professor, Department of Kinesiology

Following Achilles Tendon rupture and surgical repair tendon structural and material properties are changed from pre-injury values. These changes have been implicated as contributing to the lasting plantar flexion strength deficits that often occur following Achilles Tendon rupture. Musculoskeletal simulation offers an approach that can be used to assess these changes and can be used to guide surgery and rehabilitation. However, a procedure to develop these models in a subject specific manor would improve their utility. The purpose of this project will be to use ultrasound and strength testing measurements to develop a subject specific model of the plantar flexors that can be used to assess the effects of Achilles Tendon rupture and rehabilitation.  Specifically, students will: 1) collect ultrasound data from the Aixplorer ultrasound system on muscle and tendon properties; 2) input a range of parameter values into a muscle model developed in MATLAB; and 3) examine simulated functional outcomes.

Understanding Knee Joint Mechanics through Computational Modeling

Faculty Mentor: Ali Vahdati, PhD, Assistant Professor, Department of Engineering

Long bones and the adjacent soft connective tissues (i.e. cartilage, tendon, meniscus) of the human body experience complex and multi-directional loading during daily activities. To better understand the loads experienced by bones, cartilage and meniscus in the knee joint, subject-specific computational models can be utilized to model tissue stress/strain distributions. The goal of this project this to utilize image segmentation software and finite element modeling techniques to develop 3D subject specific models of the human knee joint. These models can help us gain more insight into mechanics of the knee joint.

Drug delivery from hydrogel composite materials

Faculty Mentor: Michelle, Oyen, PhD, Associate Professor, Department of Engineering

The hydrophilic polymers known as hydrogels make good vehicles for drug delivery in vivo due to their intrinsic biocompatibility.  For a hydrogel to be an effective carrier it must have good transport properties as well as intrinsic stiffness to withstand forces in the body over an extended duration for controlled release.  In this project, both experimental and computational approaches will be used to consider how the addition of a particulate second phase to stiffen hydrogels affects the transport properties of hydrogel composites in drug delivery applications.  A model protein (bovine serum albumin) will be used as the drug and a range of hydrogel composites of different base chemistries will be synthesized and characterized for mechanical properties and drug elution profiles.

Non-Invasive 3D Modeling and Visualization of Biological Systems

Faculty Mentor: Zhen Zhu, PhD, Assistant Professor, Department of Engineering

Various types of sensors and imaging devices have been developed for visualization, clinical analysis and noninvasive diagnosis. For example, CT, MRI and ultrasound can all be used to produce 2D or 3D scans. Preliminary results from previous REU student research projects have shown that real-time 3D modeling and visualization is indeed feasible by using existing software libraries and embedded hardware, based on which we are now able to develop biomedical applications of virtual reality and augmented reality.  The goal of this REU project is to develop a virtual reality or augmented reality system for a generic biological system, such as an organ. To accomplish these purposes, the student will: 1) study existing algorithms and software libraries associated with a 3D goggle; 2) optimize them for 3D modeling and visualization of biological systems; 3) implement them in real-time software and 4) evaluate the real-time performance.

Bone Loading during Physical Activity

Faculty Mentor: Stacey Meardon, PhD, PT, Assistant Professor, Department of Physical Therapy

Optimal loading patterns of bone is postulated to enhance adaptation and distribute stress across tissues during physical activity. Musculoskeletal modeling offers a non-invasive way in which to estimate tissue loads during everyday activity. To adequately capture the bone loading environment non-invasively, a 3D finite element (FE) model fully representing the structure of the tissue of interest is needed. Activity-specific estimates of loads experienced during activity can be incorporated in to FE analysis, allowing for greater understanding of the interaction between structure and function. Subject-specific imaging, motion capture, and subsequent processing are needed for input to such models. The aim of this project is to examine the influence of loading patterns in bone stress and strain. To accomplish this aim and in collaboration with the research team, the student will: 1) Collect and analyze motion capture data and necessary inputs for FE analysis using Qualisys, V3D, and MATLAB, 2) Employ image processing techniques to create a FE model of the tibia, a common site of injury in active populations, using image processing software, and 3) Analyze the influence of loading patterns on tibial stress, strain and failure using appropriate statistical methods.

Modeling of Cortical Dynamical Networks in Motor Control

Faculty Mentor: Chris Mizelle, PhD, Assistant Professor, Department of Kinesiology

Left-handed individuals are often overlooked in the motor control literature for various reasons. It is well known that right-handed individuals activate networks in the left side of the brain while performing motor acts with their dominant right hand. Traditionally, it has been assumed that left-handed individuals would show brain activations in the right hemisphere that “mirrored” their right-handed counterparts. However, new research has caused some doubt in this assumption. To better understand the motor neurophysiology in left-handed individuals, and how these mechanisms differ from right-handed individuals, direct study of neural activations is needed in left-handed individuals. Traditional measures of brain activations acquired using electroencephalography (EEG) are focused on evaluation of time-voltage and time-frequency analyses of sensor-level data. These are useful techniques for developing an understanding of general brain activations. Recent computational and mathematical developments, however, provide researchers with more advanced techniques to study the localization of brain activations as well as information flow dynamics using EEG signals. For this project, EEG will be used to image neural activations and a neural networks model will be developed to describe the information flow from one sensor to other sensors in response to a particular task or stimulus. Source localization methods will then be applied to isolate the brain regions responsible for generating the observed activations. Past BME-SIM REU students conducted feasibility studies, which have shown that information flow measures are sensitive to different motor control conditions, and the purpose of this project is to extend this work using EEG and measures of information flow, as well as source localization, to determine the cortical networks active in left- and right-handed individuals in different motor and cognitive-motor tasks. To accomplish this purpose, the student will: 1) become familiar with MATLAB for implementation of information flow measures; 2) become proficient at EEG data acquisition; 3) validate the neural network model by evaluating information flow measures in different behavioral domains; 4) compare estimated information flow dynamics between left- and right-handed individuals; and 5) compare source localization results between left- and right-handed individuals.

Visual Motor Control as Indicator of Brain Injury and Neurological Function

Faculty Mentor: Nick Murray, PhD, Associate Professor, Department of Kinesiology

When evaluating products consumers take a variety of factors into consideration. In addition to aesthetics and functionality, how a product makes consumers feel is an increasingly important concern. As of late, there has been a fair amount of work exploring the role of emotions in product design. However, much of this work has been more qualitative in nature than quantitative. As such, there are open questions about how to quantify emotional response to products and how the data can be used to design products that are more emotionally appealing. Neural imaging provides one approach to quantifying this data. Participants in this study will make product judgements while being monitored by electroencephalography (EEG). The goal is to uncover any insights from the neurological activity data collected during product elections that can be used to improve the performance of the emotion based utility models. Specifically, students will: 1) collect neurophysiological data (EEG) from healthy young as they evaluate consumer products; 2) process EEG data in the time and frequency domains; 3) calculate event-related potentials and event-related synchronization and desynchronization; and 4) estimate underlying neuroanatomical generators of the EEG waveforms;

Neuroimaging of physiological response during the completion of concurrent cognitive tasks in the presence of optical flow simulation

Faculty Mentor: Dr. Brian Sylcott, Assistant Professor, Department of Engineering

Vehicle operators can encounter a series of physical and cognitive challenges. For example, one of the more pressing considerations during space travel is spatial disorientation. In humans, the vestibular system is a set of sensors that allows us to maintain a sense of balance. This system is responsible for providing a sense of position and motion in three-dimensional space. The spatial disorientation astronauts encounter may be due to physiological limitations of the vestibular system. However, little is known about how the human brain can misinterpret sensory information during the sensory integration process. In addition, there is limited information available about how the human brain performs cognitive tasks while concurrently interpreting large amounts of information received during constant vestibular stimulation. In order to gain greater insight into such situations, this research will employ a novel neuroimaging technology called functional near-infrared spectroscopy (fNIRS) and eye-tracking will be used to observe physiological response during the completion of concurrent cognitive tasks in the presence of optical flow simulation in the in medial-lateral direction. Specifically, students will: 1) collect neurophysiological data from healthy adults as they complete tasks; 2) process fNIRS and eye-tracking data;

2018

Computational Fluid Dynamics Model of Coronary Artery Bypass Graft

Faculty Mentor: Dr. Stephanie George, Assistant Professor Department of Engineering

Computational Fluid Dynamics (CFD) models, combined with imaging, allow for the development of subject, or phantom specific hemodynamic models. These models may be used to diagnose, understand underlying mechanisms of disease, evaluate treatment success, evaluate and test surgical procedures, or predict clinical events or outcomes. Previous REU students have developed CFD models in the pulmonary artery and image processing codes to improve disease diagnosis. The clinical context for this project is coronary artery bypass graft (CABG) flow. Many CFD studies have examined the geometry of the anastomoses and its effect on competitive flow, however few have considered the downstream effects of perfusion/resistance and collateral flow.  A previous REU student investigated the impact of collateral flow on CABG.  The goal of this REU project is to determine the effect of downstream resistance on flow in a CABG CFD model including validation with iCertainty™. This project will utilize a phantom model of the coronary circulation including an adjustable stenosis and downstream resistance, peristaltic pump and novel optical imaging method, iCertainty™. To accomplish these purposes, the student will: 1) process CT images to obtain the geometry of the phantom model; 2) mesh the fluid volume using ANSYS Workbench; 3) develop the CFD cases (ANSYS FLUENT) with appropriate boundary conditions from iCertainty™; 4) visualize results using ANSYS Post-Processing; and 5) determine the impact of downstream resistance on CABG flow.

Effect of Semitendinosus Tendon Function Post-Harvest on Muscle Strength

Faculty Mentor: Dr. Zachary Domire, Associate Professor, Department of Kinesiology

Hamstring tendon autografts, specifically the semitendinosus, for ACL reconstruction is common and has shown excellent outcomes. However, the use of the hamstrings as a source of donor tissue is potentially problematic as the hamstrings are protective of the ACL.  While there is considerable evidence that the tendon “regenerates” following tissue harvesting, our preliminary evidence suggests that the new tendon tissue is far more compliant that the native tissue. The purpose of this project will be to test the effect of this tissue change on muscle function using a muscle model. Specifically students will: 1) collect ultrasound data from the Aixplorer ultrasound system on the tendon properties; 2) input a range of parameter values into a muscle model developed in MATLAB; 3) examine functional outcomes; and 4) compare results with experimentally measured strength data.

Effect of Altered Semitendinosus Regeneration Post-Harvest on Muscle Moment Arm and Function

Faculty Mentor: Anthony Kulas, PhD, Associate Professor, Dept. of Health Education & Promotion

The semitendinosus tendon is a common graft source used to reconstruct the ACL.  There is considerable evidence in the literature that following tendon harvesting there is tissue regeneration. However, there is also evidence that the attachment site for the new tendon may be altered. The effect of the altered attachment site on muscle moment arm and function is not fully understood. Given, the inherent protective effect the hamstring muscles have on the ACL, understanding the effects of tissue regeneration is critical towards improving outcomes post-ACL reconstruction.  The current project will develop a musculoskeletal model of the semitendinosus and examine the effects of varying insertion sites on muscle moment arm.  Furthermore, the effects on muscle function due to the altered moment arm will be examined. Specifically students will: 1) read and analyze literature on altered insertion sites in native and harvested semitendinosus tendons; 2) develop a musculoskeletal model in SIMM; 3) export moment arms for a range of insertion sites; 4) input altered moment arms into a muscle model developed in MATLAB; and 5) examine the effects on strength throughout the range of motion.

Engineering Novel Protein Mechanical Properties

Faculty Mentor: Nathan Hudson, Assistant Professor, Department of Physics

Many proteins perform a mechanical task, meaning they must perform a biological function in response to or in spite of external mechanical forces.  These proteins are often specialized to work properly under these conditions.  Understanding the molecular origins of these mechanical properties can give deeper insight into biology, provides novel therapeutic avenues, and can lead to the design of novel materials through reverse engineering.  In this project we will apply rational design strategies and/or directed evolution approaches to alter the mechanical properties of proteins, particularly proteins involved in blood clotting.

Non-Invasive 3D Modeling and Visualization of Biological Systems

Faculty Mentor: Zhen Zhu, PhD, Assistant Professor, Department of Engineering

Various types of sensors and imaging devices have been developed for visualization, clinical analysis and noninvasive diagnosis. For example, CT, MRI and ultrasound can all be used to produce 2D or 3D scans. Preliminary results from previous REU student research projects have shown that real-time 3D modeling and visualization is indeed feasible by using existing software libraries and embedded hardware, based on which we are now able to develop biomedical applications of virtual reality and augmented reality.

The goal of this REU project is to develop a virtual reality or augmented reality system for a generic biological system, such as an organ. To accomplish these purposes, the student will: 1) study existing algorithms and software libraries associated with a 3D goggle; 2) optimize them for 3D modeling and visualization of biological systems; 3) implement them in real-time software and 4) evaluate the real-time performance.

Cardiac Myocyte and Stem Cell Co-Cultures: A Modeling Study of the Cardiac Microenvironment and Cell-to-Cell Communication

Faculty Mentor: Barbara Muller-Borer, PhD, Professor, Department of Engineering

Stem cell transplantation is proposed as a therapeutic approach to repair or augment the function of impaired heart tissue. Understanding the mechanisms that safely and efficiently induce differentiation of an adult-derived stem cell into a cardiac myocyte is important for integrating stem cells from variable sources which enhances their use in cell therapy. This project focuses on 1) tissue scaffold design and modeling the cellular microenvironment (continuation of current REU project) [32] and 2) the use of  a two dimensional (2D) cell culture model to evaluate electrical, mechanical and ionic signals, between a modeled cardiac myocytes and adult-derived stem cells (new project). The structural and cellular models will be useful in studying the cardiac microenvironment, the relationship of cellular coupling, and factors determining molecular pathways for the regulation of transcription factors expressed in cardiac myocytes and upregulated in differentiating stem cells. REU students will be introduced to scaffold design and electrospinning techniques, cell culture techniques, confocal fluorescence imaging and time-lapse studies of live cell cultures, and the use of Virtual Cell Modeling and Analysis Software [33] and MATLAB. Specifically, the student will: 1) Use confocal or SEM images of biological scaffolds (results provided from current REU project) to create an analytical model to evaluate mechanical and fluid flow properties of the microenvironment; 2) simulate the spatial and temporal characteristics of the cycling cardiac cytosolic calcium signal and corresponding calcium signal in an adjacent stem cell; 3) predict/analyze the role of calcium diffusion through gap junction mediated cell-to-cell communication and other membrane channels; and 4) assess GJ mediated intercellular communication as the permeability and conductance of myocardial GJ channels are metabolically altered.

Modeling of Cortical Dynamical Networks in Motor Control

Faculty Mentor: Chris Mizelle, PhD, Assistant Professor, Department of Kinesiology

Left-handed individuals are often overlooked in the motor control literature for various reasons. It is well known that right-handed individuals activate networks in the left side of the brain while performing motor acts with their dominant right hand. Traditionally, it has been assumed that left-handed individuals would show brain activations in the right hemisphere that “mirrored” their right-handed counterparts. However, new research has caused some doubt in this assumption. To better understand the motor neurophysiology in left-handed individuals, and how these mechanisms differ from right-handed individuals, direct study of neural activations is needed in left-handed individuals. Traditional measures of brain activations acquired using electroencephalography (EEG) are focused on evaluation of time-voltage and time-frequency analyses of sensor-level data. These are useful techniques for developing an understanding of general brain activations. Recent computational and mathematical developments, however, provide researchers with more advanced techniques to study the localization of brain activations as well as information flow dynamics using EEG signals. For this project, EEG will be used to image neural activations and a neural networks model will be developed to describe the information flow from one sensor to other sensors in response to a particular task or stimulus. Source localization methods will then be applied to isolate the brain regions responsible for generating the observed activations. Past BME-SIM REU students conducted feasibility studies, which have shown that information flow measures are sensitive to different motor control conditions, and the purpose of this project is to extend this work using EEG and measures of information flow, as well as source localization, to determine the cortical networks active in left- and right-handed individuals in different motor and cognitive-motor tasks. To accomplish this purpose, the student will: 1) become familiar with MATLAB for implementation of information flow measures; 2) become proficient at EEG data acquisition; 3) validate the neural network model by evaluating information flow measures in different behavioral domains; 4) compare estimated information flow dynamics between left- and right-handed individuals; and 5) compare source localization results between left- and right-handed individuals.

Improving Visualization and Image Processing of Dynamic MRI Data for Studying the Velopharyngeal Motions during Speech

Faculty Mentor: Jamie Perry, PhD, Associate Professor, Department of Communication Sciences and Disorders

Speech is a complex system of study because it involves the movement of multiple articulators, is non-rhythmic, and motions occur in multiple planes. The velopharyngeal system is the region of speech that is important in producing normal sounding speech, that is, it has a balance of oral and nasal resonance. When this system is dysfunctional, we can perceive a resonance disorder such as hypernasality. Dynamic MRI has been proposed and used as a method of assessing the complex movements for velopharyngeal function and it is beneficial in that it provides no ionizing radiation, is non-invasive, and can accurately calculate the in-plane orifice size and velopharyngeal gap for quantitative analyses (Perry et al., 2014; Kuehn et al., 2004). Additionally, dynamic MRI has been shown to be successful in young children (Perry et al., 2016; Kollara et al., 2015). A significant obstacle of this imaging method, however is the large volume of data and lack of image processing methods to analyze the data into meaningful outputs that can inform the examiner. The current project will 1) explore methods for providing visual and numerical outputs to quantify and assess velar movements, 2) establish computational models that can calculate impact of velar muscle contraction rates (based on resistance, force, and leverage of the musculature and soft tissue structures, and 3) explore how image processing can be used to inform the clinical process in cleft palate care.

Edema Classification Based-on Statistical Traits from Image Processing

Faculty Mentor: Jason Yao, PhD, Professor, Department of Engineering

Assessing extremity edema can provide insight about a patient’s heart condition. In current clinical practices, edema is scored by healthcare professionals using a skin pressing test, where the rate at which the skin bounces back after the press is observed and evaluated. This method is subjective and often inconsistent.

From an earlier research project, a device has been developed, which uses compress air to mimic the skin pressing test and uses cameras to observe the skin re-bounce. Images captured from this process are analyzed to find the skin edema.

The purpose of this REU project is to develop statistical traits from these images to classify peripheral edema. Research work will include: observe the compress air generated skin indenting process, evaluate images from earlier experiments, develop statistical traits, and test their classification effectiveness. The extract of the traits and classification algorithm will be written in MATLAB. If developed successfully, these traits will be adopted into future automated edema assessment device to provide objective and consistent edema scoring.

Visual Motor Control as Indicator of Brain Injury and Neurological Function

Faculty Mentor: Nick Murray, PhD, Associate Professor, Department of Kinesiology

When evaluating products consumers take a variety of factors into consideration. In addition to aesthetics and functionality, how a product makes consumers feel is an increasingly important concern. As of late, there has been a fair amount of work exploring the role of emotions in product design. However, much of this work has been more qualitative in nature than quantitative. As such, there are open questions about how to quantify emotional response to products and how the data can be used to design products that are more emotionally appealing. Neural imaging provides one approach to quantifying this data. Participants in this study will make product judgements while being monitored by electroencephalography (EEG). The goal is to uncover any insights from the neurological activity data collected during product elections that can be used to improve the performance of the emotion based utility models. Specifically, students will: 1) collect neurophysiological data (EEG) from healthy young as they evaluate consumer products; 2) process EEG data in the time and frequency domains; 3) calculate event-related potentials and event-related synchronization and desynchronization; and 4) estimate underlying neuroanatomical generators of the EEG waveforms;

2017

Computational Fluid Dynamics Model of Coronary Artery Bypass Graft Competitive Flow

Faculty Mentor: Dr. Stephanie George, Department of Engineering

Computational Fluid Dynamics (CFD) models, combined with imaging, allow for the development of subject, or phantom specific hemodynamic models. These models may be used to diagnose, understand underlying mechanisms of disease, evaluate treatment success, evaluate and test surgical procedures, or predict clinical events or outcomes. Previous REU students have developed CFD models in the pulmonary artery and image processing codes to improve disease diagnosis. The clinical context for this project is coronary artery bypass graft (CABG) flow. Many CFD studies have examined the geometry of the anastomoses and its effect on competitive flow, however few have considered the downstream effects of perfusion and collateral flow. Competitive flow is defined as the influence of the parent coronary artery on the flow in the artery graft, which is not measureable at surgery. This project will utilize a phantom model of the coronary circulation including an adjustable stenosis, peristaltic pump and novel optical imaging method, iCertainty™. The goal of this REU project is to determine the effect of collateral flow on competitive flow in a CABG CFD model at varying levels of stenosis. To accomplish these purposes, the student will: 1) create geometries and computational meshes for CABG models with 50%, 75% and 90% stenosis using ANSYS Workbench; 2) develop the CFD cases (ANSYS FLUENT) for the three geometries with appropriate boundary conditions from iCertainty™; 3) Incorporate downstream collateral flow into the models and run the CFD cases; 4) compare the results using TECPLOT; and 5) determine the impact of collateral flow on CABG flow.

Improved Modeling of Muscle Fiber Length Changes In Biomechanical Models

Faculty Mentor: Dr. Zachary Domire, Department of Kinesiology

Most biomechanical models assume that muscle fibers shorten by a factor equal to the overall change in muscle length multiplied by the cosine of the muscle pennation angle. However, there is considerable evidence that muscle fibers shorten by significantly less than this. This is likely because some muscle fibers operate in series rather than in parallel. There are important implications for this assumption in both the force-length and force-velocity relationships of the muscle. The purpose of this project will be to use ultrasound imaging to measure muscle moment arms to predict whole muscle change in length and measure muscle fascicle length changes to develop a better way to represent this in future muscle models. It is hypothesized that this discrepancy can be simply incorporated into a muscle and that doing so can improve model accuracy. The model will be validated by comparing output to experimental measured strength curves. For this project, the student will: 1) collect ultrasound data from the Aixplorer ultrasound system on the lower extremity musculature; 2) calculate muscle moment arms and fascicle lengths; 3) measure isometric strength throughout the range of motion; 4) input parameters into a muscle model developed in MATLAB; and 5) compare model results with experimentally measured strength with various representations of fascicle length changes.

Use of a Subject Specific Finite Element Model to Examine the Influence of Movement Variability on Stress Volume in Bone During Physical Activity

Faculty Mentor: Stacey A. Meardon, Assistant Professor, Department of Physical Therapy

Movement variability is proposed to be necessary for adaptation [26] and stress distribution across tissues during active tasks [27]. Low movement variability has the potential to produce areas of concentrated tissue stress which may contribute to micro damage accumulation and injury [27]. However, this has not been empirically examined. In order to adequately capture stress volume, a 3D finite element model fully representing the structure of the tissue of interest is needed. Subject-specific imaging and subsequent processing are needed for input to such models [28,29]. The aim of this study is to examine the influence of coordinative variability of key lower extremity segments on bone stress volume of the tibia using a subject-specific finite element model. The working hypothesis is that reduced coordinative variability of the lower extremity will result in elevated tibial stress measures increasing the likelihood of micro damage accumulation. To accomplish this purpose using a database of MRI images of tibiae, the student will: 1) identify key coordinative and bone stress variables from the literature to guide analysis; 2) process images from existing database using image processing tools in MATLAB; 3) implement a protocol for finite element analysis of tibiae in collaboration with research team; and 4) perform statistical analysis, with guidance of research team to identify the relationship between movement variability and bone stress.

Real-Time Image Processing and Sensor Integration for Non-Invasive 3D Modeling

Faculty Mentor: Zhen Zhu, PhD, Assistant Professor, Department of Engineering

Various types of sensors and imaging devices have been developed for visualization, clinical analysis and noninvasive diagnosis. For example, CT, MRI and ultrasound can all be used to produce 2D or 3D scans. Infrared cameras have been used to detect light and heat images, which could also be used in 3D modeling. However, imagery of biomedical systems often suffers from noise, interference, blurriness or unwanted motion/vibration caused by the environment, the patient/target, or the user. Image processing techniques have to be customized and reconfigured for this type of imagery [46,47]. Preliminary results from previous REU student research projects have shown that real-time image processing and 3D modeling is indeed feasible by using open-source software libraries. Imagery data can be combined with input from additional sensors, such as motion and displacement, to create more accurate 2D or 3D models. The goal of this REU project is to identify and evaluate the sensor integration algorithms that can be implemented for real-time 3D modeling. To accomplish these purposes, the student will: 1) compare existing software libraries and identify candidate algorithms and libraries; 2) optimize them for biomedical systems; 3) implement them in real-time software; and 4) compare the performance across different algorithms, libraries and computational hardware.

Cardiac Myocyte and Stem Cell Co-Cultures: A Modeling Study of the Cardiac Microenvironment and Cell-to-Cell Communication

Faculty Mentor: Barbara Muller-Borer, PhD, Department of Engineering

Stem cell transplantation is proposed as a therapeutic approach to repair or augment the function of impaired heart tissue. Understanding the mechanisms that safely and efficiently induce differentiation of an adult-derived stem cell into a cardiac myocyte is important for integrating stem cells from variable sources which enhances their use in cell therapy. This project focuses on 1) tissue scaffold design and modeling the cellular microenvironment (continuation of current REU project) [32] and 2) the use of  a two dimensional (2D) cell culture model to evaluate electrical, mechanical and ionic signals, between a modeled cardiac myocytes and adult-derived stem cells (new project). The structural and cellular models will be useful in studying the cardiac microenvironment, the relationship of cellular coupling, and factors determining molecular pathways for the regulation of transcription factors expressed in cardiac myocytes and upregulated in differentiating stem cells. REU students will be introduced to scaffold design and electrospinning techniques, cell culture techniques, confocal fluorescence imaging and time-lapse studies of live cell cultures, and the use of Virtual Cell Modeling and Analysis Software [33] and MATLAB. Specifically, the student will: 1) Use confocal or SEM images of biological scaffolds (results provided from current REU project) to create an analytical model to evaluate mechanical and fluid flow properties of the microenvironment; 2) simulate the spatial and temporal characteristics of the cycling cardiac cytosolic calcium signal and corresponding calcium signal in an adjacent stem cell; 3) predict/analyze the role of calcium diffusion through gap junction mediated cell-to-cell communication and other membrane channels; and 4) assess GJ mediated intercellular communication as the permeability and conductance of myocardial GJ channels are metabolically altered.

Development and Validation of Musculoskeletal Models to Predict Quadriceps Moments

Faculty Mentor: Anthony Kulas, PhD, Associate Professor, Dept. of Health Education & Promotion

While musculoskeletal models are commonly used to better understand mechanisms of human movement, current musculoskeletal models do not accurately predict quadriceps isometric moments through a full range of motion. Developing subject-specific models that accurately predict muscle strength among healthy individuals is a critical first step towards ultimately understanding mechanisms responsible for both strength gains (performance enhancement) and strength loss (disease states). The purpose of this project will be to develop subject-specific models to predict isometric quadriceps torque profiles reflective of a healthy and young population. Specifically the models will: 1) incorporate the relationship between vastus lateralis fascicle lengthening to whole muscle lengthening and 2) be reflective of in-vivo patellofemoral motion. Comparison to experimental isometric quadriceps torque curves will serve as the criterion for model validation. Specifically, the student will: 1) determine the relationship between vastus lateralis (a surrogate of quadriceps function) fascicle lengthening to muscle lengthening, 2) adapt the model patellofemoral motion to reflect in-vivo motion, and 3) compare model predicted quadriceps torque profiles to experimentally measured quadriceps torques.

Modeling of Cortical Dynamical Networks in Motor Control

Faculty Mentor: Chris Mizelle, PhD, Assistant Professor, Department of Kinesiology

Left-handed individuals are often overlooked in the motor control literature for various reasons. It is well known that right-handed individuals activate networks in the left side of the brain while performing motor acts with their dominant right hand. Traditionally, it has been assumed that left-handed individuals would show brain activations in the right hemisphere that “mirrored” their right-handed counterparts [30]. However, new research has caused some doubt in this assumption [31]. To better understand the motor neurophysiology in left-handed individuals, and how these mechanisms differ from right-handed individuals, direct study of neural activations is needed in left-handed individuals. Electroencephalography (EEG) will be used to image neural activations and a neural networks model will be developed to describe the information flow from one sensor to other sensors in response to a particular task or stimulus. A past BME-SIM REU student conducted a feasibility study, which has shown that information flow measures are sensitive to different motor control conditions, and the purpose of this project is to extend this work using EEG and measures of information flow to determine the cortical networks active in left and right handed individual in different motor and cognitive-motor tasks. To accomplish this purpose, the student will: 1) become familiar with MATLAB for implementation of information flow measures; 2) become proficient at EEG data acquisition; 3) validate the neural network model by evaluating information flow measures in different behavioral domains; and 4) compare estimated information flow dynamics between left- and right-handed individuals.

Numerical Modeling of a Multi-Degree of Freedom Resonant System

Faculty Mentor: Teresa Ryan, PhD, Assistant Professor, Department of Engineering

The student will contribute to ongoing work in the use of coupled arrays of resonant structures [34-39] for a variety of applications. This topic has roots in a simple single degree of freedom dynamic vibration absorber, which pulls energy at a certain single frequency away from a primary resonator. Work has expanded to include attachment of a set of resonant structures, termed a subordinate oscillator array.  Such complex vibratory systems have possible applications as mechanical filters or absorbers of vibration energy (safety and injury prevention), energy harvesting (such as battery charging based on gait), or ultrasensitive mass detection (breathalyzer for new analytes) [35-38]. The dynamic response of any resonant system can be manipulated by attaching a subordinate oscillator array with a prescribed distribution of properties. Rapid prototyping technology can be used for quick production of design iterations, but the anisotropic nature of the resulting structures requires careful evaluation. Ultimately, variation of mass and stiffness dictate behavior of the vibrating system which has been shown to be highly sensitive to minute variation in those property distributions. The student will design and build a set of resonator arrays designed to trap energy in a chosen frequency range. Specifically, the student will: 1) use different rapid prototyping technologies and/or build parameters to fabricate the arrays; 2) Use the dynamic material property measurements of the prototypes in a MATLAB based numerical model of the multi-degree of freedom resonant system; and 3) Use the model to determine the sensitivity of array performance to the different build parameters.

Predictors of Knee Joint Loads in Army Cadets During Marching with Body-Borne Loads

Faculty Mentor: Richard Willy, Assistant Professor, Department of Physical Therapy

Soldier mobility and performance is often impaired due to lower extremity injury resulting from high peak and cumulative load demands [40]. In soldiers, the knee is highly susceptible to injury, particularly while marching with body-borne loads [41]. Therefore, algorithms to predict knee loads in soldiers, both with and without body-borne loads, are highly sought. While various screening tools have attempted to predict lower extremity loads and subsequent injury risk in soldiers, these assessments have proven to either lack precision or are not feasible for wholesale adoption in the United States Military. As such, this proposed project aims to develop field-ready algorithms that can accurately predict knee joint loads in soldiers while marching with and without body-borne loads. To accomplish these purposes, the student will: 1) will estimate tibiofemoral and patellofemoral contact forces during marching with and without 50 pounds of body-borne load in a cohort of university-based, Army Reserve Officer Training Corps Cadets via a previously described musculoskeletal model [42]; 2) measure various clinical assessments of lower extremity strength and flexibility as well as aerobic fitness; 3) develop a regression equation to estimate tibiofemoral and patellofemoral contact forces using measurements obtained from the clinical assessments.

Quantifying Consumer Emotional Responses using EEG and Matlab

Faculty Mentor: Brian Sylcott, PhD, Assistant Professor, Department of Engineering

When evaluating products consumers take a variety of factors into consideration. In addition to aesthetics and functionality, how a product makes consumers feel is an increasingly important concern. As of late, there has been a fair amount of work exploring the role of emotions in product design. However, much of this work has been more qualitative in nature than quantitative. As such, there are open questions about how to quantify emotional response to products and how the data can be used to design products that are more emotionally appealing. Neural imaging provides one approach to quantifying this data. Participants in this study will make product judgements while being monitored by electroencephalography (EEG). The goal is to uncover any insights from the neurological activity data collected during product elections that can be used to improve the performance of the emotion based utility models. Specifically, students will: 1) collect neurophysiological data (EEG) from healthy young as they evaluate consumer products; 2) process EEG data in the time and frequency domains; 3) calculate event-related potentials and event-related synchronization and desynchronization; and 4) estimate underlying neuroanatomical generators of the EEG waveforms;