Patients with chronic lower back pain (CLBP) typically demonstrate different biomechanics than healthy controls during a lifting task. Motion differences in a repetitive lifting task have been described previously using differences in the timing of body angles changes during the lift. These timing changes rely on small differences of motion and are difficult to measure and to interpret. The purpose of this study is to evaluate shoulder jerk (rate of change of acceleration) in a repetitive lifting task as a parameter to detect differences of motion between controls and CLBP patients and to measure the impact of a rehabilitation program on jerk. The jerk calculation proved to be a noisy measure since jerk is the third derivative of position, and a simulation study was performed to evaluate smoothing methods to provide the best estimates of the third derivative. Woltrings generalized cross-validation spline produced the best estimates and was fit to subject data. Derivatives were calculated using differentiation of the spline coefficients, and root-means-square (rms) amplitude of jerk was used for comparison. Lifts were divided into phases of early, middle or late based on the number of repetitions completed by the subject. Average values of rms jerk during a lift were computed at each of the task phases. Significant group differences were found for rms jerk. CLBP patients were found to perform lifts with lower jerk values than controls and as the task progressed, rms jerk increased for both groups. A group-by-phase interaction was significant. After completion of a rehabilitation program, CLBP patients performed lifts with greater rms jerk. In general, patients performed lifts with lower jerk values than controls, suggesting that pain impacts lifting style.

Key to advancing stem cell utilization in regenerative medicine and cell-based therapies is the development of systems to expand cells to clinically relevant numbers while maintaining the desired stem cell phenotype. Mathematical growth models play an important role in developing standardized systems, as they are both predictive tools for expansion potential and tools to describe current kinetic parameters of a stem cell population. One disease that may benefit from cell therapy is Duchenne Muscular Dystrophy (DMD), a muscle disease characterized by the lack of dystrophin expression at the sarcolemma of muscle fibers resulting in muscle fiber necrosis and muscle weakness. While transplantation of normal myoblasts into dystrophin-deficient muscle can restore dystrophin, the use of muscle-derived stem cells (MDSC) has enhanced the success of cell transplantation. For these reasons, muscle stem cell isolation and the development of transplantation techniques have garnered increased attention recently. One limitation of MDSC use is the few numbers of cells available from a muscle biopsy, thus presenting the requirement for in vitro expansion. The overall goal of this study was to provide a thorough quantitative examination of the expansion of MDSC populations. In this project, an imaging system was established to analyze stem cell expansion. The applicability of this system was demonstrated in MDSC expansion with cytokine stimulation. It was found that accounting for the proliferative heterogeneity that exists in stem cell populations would allow for more accurate estimations of kinetic parameters. Next, a more sophisticated imaging system was used to further develop an automated system for analysis of MDSC proliferation and behavioral characterization. Finally, an understanding of the limits of expansion was explored. The role of long-term expansion on stem cell phenotype and regeneration capacity was examined to consider the issue of quantity vs. quality of muscle-derived stem cells. This study provided a systematic method for assessing expansion and an in-depth investigation into the natural progression of stem cell expansion. It is expected that these findings will provide a biological understanding of the limits of expansion and a foundation for more standardized methods of expansion of MDSC as MDSC are advanced to a clinical setting.

Injuries to knee ligaments are frequent, demanding an increased understanding of the healing process. Clinically, the injured medial collateral ligament (MCL) has been found to heal without surgical intervention. However, laboratory studies have shown that, even one year after injury, the biomechanical properties, biochemical composition, and histomorphological appearance of the healing MCL remains suboptimal. While research has focused on the changes in mechanical properties (i.e. stress-strain behavior) of the healed MCL, studies on its viscoelastic properties are limited. Yet, this knowledge is critical to determine the overall kinetic response of the knee joint. The quasi-linear viscoelastic (QLV) theory proposed by Professor Y.C. Fung has been frequently used to model the viscoelastic properties of the MCL. This theory was developed based on an idealized step-elongation during a stress relaxation test. As this is experimentally impossible, the constants of the theory may not be representative when they are determine based on experiments that utilize finite strain rates. Thus, the overall objectives of this dissertation were to 1) develop and validate a novel experimental and analytical approach that accounts for finite strain rates and provides an accurate determination of the viscoelastic properties of the normal MCL, 2) apply this new approach to describe the viscoelastic behavior of the healing MCL, and 3) to determine whether the new approach can describe the response of the MCL to harmonic oscillations. This work demonstrated that a newly developed approach could be utilized to determine the constants of the quasi-linear viscoelastic theory and successfully describe the viscoelastic behavior of both the normal and healing MCLs. Interestingly, the healing ligaments display a lower initial slope of the stress-strain curve and a greater propensity to dissipate energy, suggesting other structures within the knee would have to play a compensatory role in knee function. It was also found that the mechanisms governing the viscoelastic response of the MCL to harmonic oscillations may not be the same as that which governs stress relaxation behavior. Thus, a more general theory may be necessary to describe both phenomena.

This dissertation concerns the development of a device called the Sonic Flashlight, which employs a novel method for viewing real-time ultrasound images inside the body exactly at the location where it is being scanned. While other augmented reality methods have previously been developed to view ultrasound and other medical imaging modalities within the body, they are generally much more complicated, slower and less robust than the Sonic Flashlight. In this dissertation, we aim to develop the Sonic Flashlight towards one particular clinical application, central vascular access, and lay the groundwork leading to the first clinical trials. The goal of central vascular access is to insert a catheter into a major vein to deliver medications in large quantities. These veins are usually not visible to the naked eye, so real-time ultrasound is employed to guide the needle into them. While real-time ultrasound guidance significantly enhances the safety of central venous access, learning this skill can be a challenge for the novice user, one major obstacle being the displaced sense of hand-eye coordination that occurs when the operator must look away from the operating field to view the conventional ultrasound monitor. We developed the 5th generation Sonic Flashlight, as well as a novel calibration method, called thin-gel calibration, as part of this dissertation. The thin-gel system allows us to accurately calibrate the Sonic Flashlight and measure the calibration accuracy. Finally, experiments were conducted with a variety of subject populations using vascular ultrasound phantoms and cadavers to validate Sonic Flashlight guidance, demonstrating that the device is ready for clinical trials.

Fractures are the second most common presentation of child abuse, soft tissue injury being the most common. Femurs are the most common long-bone fractured in inflicted injury. When a child presents to the emergency department, a clinician must judge if the childs fracture matches the account provided by the caregiver. An objective tool is needed to aid in the assessment of injury plausibility. Predicting femur fracture strength is key to developing this tool. Immature porcine femurs are widely used to model pediatric human femurs. This study investigated immature porcine femur fracture load, energy to failure and stiffness in three-point bending, torsion and axial compression, with and without soft tissue intact and at different displacement rates. Significant differences exist between three point bending with soft tissue intact (n=6) and devoid of soft tissue (n=6) for stiffness (means=1607.9 lbf/in. and 1981.9 lbf/in, respectively, p=0.046) and energy to failure (means=36.9 in-lbf and 25.0 in-lbf, respectively, p=0.046). Torsion tests show significant differences in the fracture torque between groups tested at 0.167 degrees/sec (n=7) and 90 degrees/sec (n=7, means=30.69 in-lbf and 46.13 in-lbf, respectively, p=0.018). Axial compression experiments at 0.04 in/sec (n=5) resulted in fracture load, energy to failure and stiffness of 273.4 lbf, 70.7 in-lbf and 829.4 lbf/in, respectively, while axial compression experiments at 2 in/sec (n=2) resulted in higher fracture loads, energy to failure and stiffness (441 lbf, 154.2 in-lbf and 1894 lbf/in, respectively). Three-point bending tests resulted in oblique or transverse fractures, torsion and axial compression tests resulted in spiral and growth plate fractures, respectively. Correlations between bone mineral density and structure geometry showed promise as a predictive model for femur fracture response in all loading mechanisms. Multivariable regression modeling resulted in high R2 values (0.62 0.74) for femurs tested with soft tissue intact in three-point bending, but low values (0.22 0.29) for femurs tested devoid of soft tissue in three-point bending; relatively high R2 values (0.66 0.78) for fracture torque in torsion and low R2 values (0.22 0.47) for energy to failure in torsion. Further investigation with a larger sample is needed to reliably predict immature femur fracture response.

The use of a ventricular assist device (VAD) is a promising option for the treatment of end-stage heart failure. In many cases VADs provide not only temporary support, but contribute to the recovery of the native ventricle. Many studies have reported incidences where the native ventricle has recovered function, leading to device explantation and eliminating the need for heart transplantation. Despite strong interest in the subject for many years, the determinants of the recovery process are poorly understood and number of patients successfully weaned from chronic support remains low. A mathematical model was developed to gain an understanding of the complex mechanical interactions between a pneumatic, pulsatile VAD and the left ventricle. The VAD model was verified in-vitro using a mock circulatory loop. Over a wide range of experimental conditions, it correctly described observed dynamic behaviors and was accurate in predicting both VAD stroke volume and fill-to-empty rate within 6% error. This validated VAD model was coupled to a simple, lumped parameter cardiovascular model. The coupled model qualitatively reproduced the temporal patterns of various hemodynamic variables observed in clinical data. A concept of VAD characteristic frequency (fc) was developed to facilitate the analysis of VAD-ventricle synchrony. Characteristic frequency, defined as VAD rate in the absence of ventricular contraction, was essentially independent of cardiovascular parameters. For a given set of VAD parameters, synchrony was found to occur over a range of native heart rates. While the lower bound was determined by fc alone, the upper bound was a function of various cardiovascular parameters (e.g., left ventricular contractility, EMAX and systemic vascular resistance, SVR). In the case of synchronous behavior, the VAD and native heart have matched rates and counter-pulse, resulting in reduced ventricular loading. A decrease in EMAX or an increase in SVR increases asynchrony, resulting in frequent occurrences of co-pulsed beats (i.e., high ventricular loading). In conclusion, we found that VAD-ventricle synchrony is determined by a complex interaction between VAD and cardiovascular parameters. Our model-based analysis of VAD-ventricle interaction may be useful for optimizing the VAD operation, characterizing native ventricular contractility, and better understanding of the recovery process.

The reported biomechanical properties of soft tissues are often dependent on the method used to determine specimen cross-sectional area as this is an important factor in tissue stress calculations. Cross-sectional shape is also important, especially for documenting morphological changes in the tissue during healing. Successful measurement of these geometric characteristics, however, has been hindered by the complex geometry of many biological tissues, as well as concerns regarding the deformability of these tissues under mechanical contact. The overall objective of this thesis was to evaluate the cross-sectional shape and area of normal and healing ligaments using laser-based devices. Lasers allow for measurements without inducing mechanical contact and deformation. Initially, the effects of treatment with a bio-scaffold, small intestinal submucosa (SIS), on the cross-sectional shape and area of healing ligaments were evaluated using the laser micrometer system. However, due to limitations of currently available methods such as the inability to detect concavities, a new system was also developed and evaluated. The cross-sectional shape and area of non-treated healing and SIS-treated healing rabbit MCLs were assessed using a laser micrometer system 26 weeks after gap injury. No significant changes in shape were detected between SIS-treated and non-treated MCLs. However, SIS-treatment significantly reduced the cross-sectional area at 26 weeks after injury in comparison to the non-treated group. A charge coupled device (CCD) laser reflectance system was developed in order to determine the cross-sectional shape and area of tissues containing surface concavities. For this system, a CCD laser displacement sensor recorded distance measurements off a specimen while it rotated 360°. The area and shape could then be determined from this data. The system was evaluated using geometric shapes of known cross-sectional area before being applied to biological specimens. This work demonstrated that cross-sectional shape and area measurements can be used to quantify tissue healing and remodeling. Additionally, the CCD laser reflectance system successfully detected concavities on the surfaces of tissues and therefore is a viable approach to biological tissue measurement. However, this system does not offer much improvement in accuracy over the laser micrometer system for tissues that do not have significant concavities.

When ligaments such as the medial collateral ligament (MCL) are injured, they generally heal but form scar tissue, which is composed of a disorganized collagen matrix that is over-produced by fibroblasts. Scar tissue has inferior structural and mechanical properties, which can lead to joint instability. Excessive fibroblast contraction is thought to contribute to tissue scarring. Previous studies have shown that both TGF-â1 and TGF-â3 increase fibroblast contraction and collagen synthesis. However, TGF-â1 enhances scar tissue formation whereas TGF-â3 actually reduces it. In addition, both TGF-â isoforms have been found to increase the expression of á-SMA, which correlates with increased fibroblast contractility. An increase in tension at the wound site has also been found to increase á-SMA protein levels. Therefore these factors are all important in the wound healing process. The overall objective of this thesis research was to investigate cellular and molecular mechanisms that affect scar tissue formation in healing ligaments. Contraction forces, collagen synthesis, and á-SMA protein expression of healing and normal MCL fibroblasts in response to treatment with TGF-â1, TGF-â3, and collagen gel tension were investigated. A novel culture force monitor (CFM) system was used to quantify forces of fibroblast contraction. It was found that healing MCL fibroblasts produced greater contractile forces and higher levels of collagen synthesis than normal MCL fibroblasts. In addition, treatment with TGF-â1 or TGF-â3 increased contraction forces in healing fibroblasts compared to untreated controls, with TGF-â1 consistently producing greater contraction forces than treatment with TGF-â3. TGF-â1 and TGF-â3 also induced higher levels of á-SMA protein expression compared to untreated fibroblasts. Consistent with the contraction forces, fibroblasts treated with TGF-â1 expressed higher levels of á-SMA protein than those treated with TGF-â3. Further, it was found that when tension in gels embedded with normal MCL fibroblasts was released, expression of á-SMA protein also decreased. Thus, this study showed that healing and normal fibroblasts have differential contractile and collagen synthesis abilities. The results of this study showed that the presence of TGF-â1, TGF-â3, and tension in the matrix should be regulated to improve ligament healing. Decreasing the ratio of TGF-â1 to TGF-â3 in an injured ligament may decrease fibroblast contraction and thus reduce scar formation in healing MCLs. Finally, reducing tension levels in healing ligaments and hence down-regulating á-SMA protein expression may also decrease ligament scarring.

This thesis concentrates on the reduction of RF field (or B1) inhomogeneity in high magnetic field MRI. B1 inhomogeneity is one of the major drawbacks in high field MRI. The non-uniformity causes regions of increased and decreased signal intensity in the images. None of existing methods can perfectly correct the non uniformity. This thesis aims to develop new methods that are practical, safe, and required no additional devices. Specifically, three-dimensional (3D) tailored RF (TRF) pulse were designed and validated in human MRI experiments at 3 Tesla. Two novel designs of 3D TRF pulses for B1 inhomogeneity reduction are presented in this thesis. Both designs are based on the small flip angle approximation. The first design is for a thick slab 3D acquisition. These pulses employ a 3D stack of spirals k-space trajectory simultaneously with B1 inhomogeneity compensated RF pulse waveforms during excitation. Another pulse design uses analytical functions as a compensated B1 inhomogeneity pulse weighting function. The k-space is modeled in the manner such that kx-ky provides compensated spatial weighting function for quadratically varying B1 inhomogeneity patterns. The kz-direction is controlled by fast switching gradients in the fashion similar to Echo planar imaging (EPI). This design is more appropriate for 2D high resolution acquisition images. The two pulse designs show equal improvement of signal loss of approximately 30%. Long pulse length, 22 ms, and limited peak B1 are the major concern of the first design. The second design, the compensated fast kz pulses provide relatively shot pulse length only 3-5 ms. The primary limitation of this design is that it can be used for only a quadratic pattern of B1 inhomogeneity and may cause resonance shift.

Aortic heart valves and their replacements fail in vivo for reasons that are not fully understood. Mechanical evaluation and simulations of the function of native aortic valves and their replacements have been limited to tensile and biaxial tests that seek to quantify the behavior of leaflet tissues as a homogenous whole. However, it is widely understood that valvular tissues are multi-layered structures composed of collagen, elastin, and glycosaminoglycans. The mechanical behavior of these layers within intact valve leaflet tissues and their interactions are unknown. In addition, pulmonary valves have been used as substitutes for diseased aortic valves without any real understanding of the mechanical differences between the aortic and pulmonary valves. The pulmonary valve operates in an environment significantly different than that of the aortic valve and, thus, mechanical behavioral differences between the two valve leaflets may exist. In this study, we sought to determine the mechanical properties of the porcine aortic and pulmonary valves in flexure, and to determine the mechanical relationship between the leaflet layers: the fibrosa, spongiosa, and ventricularis. This was accomplished by developing a novel flexure mechanical testing device that allowed for the determination of the flexural stiffness of the leaflet tissue was determined using Bernoulli-Euler bending. Moreover, transmural strains were quantified and used to determine the location of the neutral axis to determine if differences existed in the layer properties of the fibrosa and ventricularis. To contrast the flexural studies, biaxial experiments were also performed on the aortic and pulmonary valves to determine the mechanical differences in the tensile behavior between the two leaflets. Results indicated that the pulmonary valve is stiffer than the aortic valve in flexure but less compliant than the aortic valve in biaxial tensile tests. The interactions between the layers of the leaflets suggest an isotropic mechanical response in flexure, but do so through mechanisms that are not fully understood. For heart valve leaflet replacement therapy, this study illustrates the biomechanical differences between the aortic and pulmonary valve leaflets and emphasizes the need to fully characterize the two as separate but related entities. Understanding the interactions of microscopic structures such as collagen and elastin fibers is critical to understanding the response of the tissue as a whole and how all these elements combine to provide a functioning component of the organ system.