Colorectal cancer is the second most common cause of cancer-related deaths in the United States, with the relative 5-year survival rate for distant stage cancer being only 14%. The most common treatment for colorectal cancer is with chemotherapeutic drugs; however, the discovery of these drugs is costly, time-consuming, and often requires the use of animal models that do not yield results that translate to clinical trials. Due to these shortcomings, researchers seek to develop physiologically relevant in vitro tumor models that more accurately mimic the tumor microenvironment for cheaper and faster high-throughput drug screening. The aim of this research was to develop a colorectal cancer tumor model co-cultured with endothelial and stromal cells, followed by validation with clinically relevant chemotherapeutic agents within microfluidic devices. The first experiment consisted of a lipofection of fibroblasts to yield fluorescently tagged cells that could be later imaged using a fluorescence microscope. The next experiment consisted of a co-culture of tumor, endothelial, and fibroblast cells at varying densities in a twodimensional (2D) culture to determine the optimal plating densities that would yield quantifiable tumor and endothelial network formation. The following experiment used these optimal densities to test the effects of the chemotherapeutic agents oxaliplatin and SN38 on the tumor and endothelial cells in 2D. After the various densities and drug concentrations were tested in 2D, the model was introduced into microfluidic devices. The first experiment in the devices was similar to the first experiment plated in 2D, as it involved the establishment of optimal plating densities of all three cell types within the devices. Similarly, the goal of this experiment was to yield quantifiable tumor and endothelial network formation within the devices. The final experiment performed in this research was the introduction of oxaliplatin and SN38 to the optimized densities v of cells determined from the previous experiment, with the aim of evaluating the effects of these chemotherapeutic agents on the tumor and endothelial cells within microfluidic devices. The two experiments plated in 2D established plating densities to be tested in the devices. These experiments also showed that increasing drug concentrations resulted in reduced tumor count and size and revealed no disruption in the endothelial networks when exposed to oxaliplatin concentrations as high as 50 µM. The final two experiments in microfluidic devices revealed that endothelial network formation is not yet possible within the devices with the current protocols, but that tumor cells still showed dose-dependent responses to drug exposure as they did in 2D. Due to the lack of network formation in this device model, future work is required to allow for endothelial cell organization into networks, to further increase the physiological relevancy of this model to in vivo tumor conditions.

Statistical data suggests that increased use of evidence-based medical therapies has largely contributed to the decrease in American death rate caused by heart disease. And my studies are about two applications of magnetic resonance imaging (MRI) as a non-invasive approach in evidence-based health care research. In my first study, the achievement of a pulmonary valve replacement surgery was assessed on a patient with tetralogy of Fallot (TOF). In order to evaluate the remodeling of right ventricle, two biventricular finite element models were built up for pre-surgical images and post-surgical images. In my second study, 3D Lagrangian strains and torsion in the left ventricle of ten rats were investigated using Displacement ENcoding with Stimulated Echoes (DENSE) cardiac magnetic resonance (CMR) images. Tools written in MATLAB were developed for 2D contouring, 3D modeling, strain and torsion computations, and statistical comparison across subjects.

Many cerebral diseases are associated with abnormal cerebral hemodynamics and impaired cerebral autoregulation (CA). CA is a mechanism to maintain cerebral blood flow (CBF) stable when mean arterial pressure (MAP) fluctuates. Evaluating these abnormalities requires direct measurements of cerebral hemodynamics and MAP. Several near-infrared diffuse optical instruments have been developed in our laboratory for hemodynamic measurements including near-infrared spectroscopy (NIRS), diffuse correlation spectroscopy (DCS), hybrid NIRS/DCS, and dual-wavelength DCS flow-oximeter. We utilized these noninvasive technologies to quantify CBF and cerebral oxygenation in different populations under different physiological conditions/manipulations. A commercial finger plethysmograph was used to continuously monitor MAP. For investigating the impact of obstructive sleep apnea (OSA) on cerebral hemodynamics and CA, a portable DCS device was used to monitor relative changes of CBF (rCBF) during bilateral thigh cuff occlusion. Compared to healthy controls, smaller reductions in rCBF and MAP following cuff deflation were observed in patients with OSA, which might result from the impaired vasodilation. However, dynamic CAs quantified in time-domain (defined by rCBF drop/MAP drop) were not significantly different between the two groups. We also evaluated dynamic CA in frequency-domain, i.e., to quantify the phase shifts of low frequency oscillations (LFOs) at 0.1 Hz between cerebral hemodynamics and MAP under 3 different physiological conditions (i.e., supine resting, head-up tilt (HUT), paced breathing). To capture dynamic LFOs, a hybrid NIRS/DCS device was upgraded to achieve faster sampling rate and better signal-to-noise. We determined the best hemodynamic parameters (i.e., CBF, oxygenated and total hemoglobin concentrations) among the measured variables and optimal physiological condition (HUT) for detecting LFOs in healthy subjects. Finally, a novel dual-wavelength DCS flow-oximeter was developed to monitor cerebral hemodynamics during HUT-induced vasovagal presyncope (VVS) in healthy subjects. rCBF was found to have the best sensitivity for the assessment of VVS among the measured variables and was likely the final trigger of VVS. A threshold of ~50% rCBF decline was observed which can completely separate subjects with or without presyncope, suggesting its potential role for predicting VVS. With further development and applications, NIRS/DCS techniques are expected to have significant impacts on the evaluation of cerebral hemodynamics and autoregulation.

In this study we design a succession of three increasingly adept diffuse optical devices towards the simultaneous 3D imaging of blood flow and fluorescence contrasts in relatively deep tissues. These metrics together can provide future insights into the relationship between blood flow distributions and fluorescent or fluorescently tagged agents. A noncontact diffuse correlation tomography (ncDCT) device was firstly developed to recover flow by mechanically scanning a lens-based apparatus across the sample. The novel flow reconstruction technique and measuring boundary curvature were advanced in tandem. The establishment of CCD camera detection with a high sampling density and flow recovery by speckle contrast followed with the next instrument, termed speckle contrast diffuse correlation tomography (scDCT). In scDCT, an optical switch sequenced coherent near-infrared light into contact-based source fibers around the sample surface. A fully noncontact reflectance mode device finalized improvements by combining noncontact scDCT (nc_scDCT) and diffuse fluorescence tomography (DFT) techniques. In the combined device, a galvo-mirror directed polarized light to the sample surface. Filters and a cross polarizer in stackable tubes promoted extracting flow indices, absorption coefficients, and fluorescence concentrations (indocyanine green, ICG). The scDCT instrumentation was validated through detection of a cubical solid tissue-like phantom heterogeneity beneath a liquid phantom (background) surface where recovery of its center and dimensions agreed with the known values. The combined nc_scDCT/DFT identified both a cubical solid phantom and a tube of stepwise varying ICG concentration (absorption and fluorescence contrast). The tube imaged by nc_scDCT/DFT exhibited expected trends in absorption and fluorescence. The tube shape, orientation, and localization were recovered in general agreement with actuality. The flow heterogeneity localization was successfully extracted and its average relative flow values in agreement with previous studies. Increasing ICG concentrations induced notable disturbances in the tube region (≥ 0.25 μM/1 μM for 785 nm/830 nm) suggesting the graduating absorption (320% increase at 785 nm) introduced errors. We observe that 830 nm is lower in the ICG absorption spectrum and the correspondingly measured flow encountered less influence than 785 nm. From these results we anticipate the best practice in future studies to be utilization of a laser source with wavelength in a low region of the ICG absorption spectrum (e.g., 830 nm) or to only monitor flow prior to ICG injection or post-clearance. In addition, ncDCT was initially tested in a mouse tumor model to examine tumor size and averaged flow changes over a four-day interval. The next steps in forwarding the combined device development include the straightforward automation of data acquisition and filter rotation and applying it to in vivo tumor studies. These animal/clinical models may seek information such as simultaneous detection of tumor flow, fluorescence, and absorption contrasts or analyzing the relationship between variably sized fluorescently tagged nanoparticles and their tumor deposition relationship to flow distributions.

Statistical data from clinical studies indicate that the death rate caused by heart disease has decreased due to an increased use of evidence-based medical therapies. This includes the use of magnetic resonance imaging (MRI), which is one of the most common non-invasive approaches in evidence-based health care research. In the current work, I present 3D Lagrangian strains and torsion in the left ventricle of healthy and isoproterenol-stimulated rats, which were investigated using Displacement ENcoding with Stimulated Echoes (DENSE) cardiac magnetic resonance (CMR) imaging. With the implementation of the 12-segment model, a detailed profile of regional cardiac mechanics was reconstructed for each subject. Statistical analysis revealed that isoproterenol induced a significant change in the strains and torsion in certain regions at the mid-ventricle level. In addition, I investigated right ventricular cardiac mechanics with the methodologies developed for the left ventricle. This included a comparison of different regions within the basal and mid-ventricular regions. Despite no regional variation found in the peak circumferential strain, the peak longitudinal strain exhibited regional variation at the anterior side of the RV due to the differences in biventricular torsion, mechanism of RV free wall contraction, and fiber architecture at RV insertions. Future applications of the experimental work presented here include the construction and validation of biventricular finite element models. Specifically, the strains predicted by the models will be statistically compared with experimental strains. In addition, the results of the present study provide an essential reference of RV baseline evaluated with DENSE MRI, a highly objective technique.

Measurement of tissue hemodynamics provides vital information for the assessment of tissue viability. This thesis reports three noninvasive near-infrared diffuse optical systems for spectroscopic measurements and tomographic imaging of tissue hemodynamics in vulnerable tissues with the goal of disease diagnosis and treatment monitoring. A hybrid near-infrared spectroscopy/diffuse correlation spectroscopy (NIRS/DCS) instrument with a contact fiber-optic probe was developed and utilized for simultaneous and continuous monitoring of blood flow (BF), blood oxygenation, and oxidative metabolism in exercising gastrocnemius. Results measured by the hybrid NIRS/DCS instrument in 37 subjects (mean age: 67 ± 6) indicated that vitamin D supplement plus aerobic training improved muscle metabolic function in older population. To reduce the interference and potential infection risk on vulnerable tissues caused by the contact measurement, a noncontact diffuse correlation spectroscopy/tomography (ncDCS/ncDCT) system was then developed. The ncDCS/ncDCT system employed optical lenses to project limited numbers of sources and detectors on the tissue surface. A motor-driven noncontact probe scanned over a region of interest to collect boundary data for three dimensional (3D) tomographic imaging of blood flow distribution. The ncDCS was tested for BF measurements in mastectomy skin flaps. Nineteen (19) patients underwent mastectomy and implant-based breast reconstruction were measured before and immediately after mastectomy. The BF index after mastectomy in each patient was normalized to its baseline value before surgery to get relative BF (rBF). Since rBF values in the patients with necrosis (n = 4) were significantly lower than those without necrosis (n = 15), rBF levels can be used to predict mastectomy skin flap necrosis. The ncDCT was tested for 3D imaging of BF distributions in chronic wounds of 5 patients. Spatial variations in BF contrasts over the wounded tissues were observed, indicating the capability of ncDCT in detecting tissue hemodynamic heterogeneities. To improve temporal/spatial resolution and avoid motion artifacts due to a long mechanical scanning of ncDCT, an electron-multiplying charge-coupled device based noncontact speckle contrast diffuse correlation tomography (scDCT) was developed. Validation of scDCT was done by imaging both high and low BF contrasts in tissue-like phantoms and human forearms. In a wound imaging study using scDCT, significant lower BF values were observed in the burned areas/volumes compared to surrounding normal tissues in two patients with burn. One limitation in this study was the potential influence of other unknown tissue optical properties such as tissue absorption coefficient (µa) on BF measurements. A new algorithm was then developed to extract both µa and BF using light intensities and speckle contrasts measured by scDCT at multiple source-detector distances. The new algorithm was validated using tissue-like liquid phantoms with varied values of µa and BF index. In-vivo validation and application of the innovative scDCT technique with the new algorithm is the subject of future work.

The high prevalence and mortality of cerebrovascular disease has led to the development of several methods to measure cerebral blood flow (CBF) in vivo. One of these, arterial spin labeling (ASL), is a quantitative magnetic resonance imaging (MRI) technique with the advantage that it is completely non-invasive. The quantification of CBF using ASL requires correction for a tissue specific parameter called the brain-blood partition coefficient (BBPC). Despite regional and inter-subject variability in BBPC, the current recommended implementation of ASL uses a constant assumed value of 0.9 mL/g for all regions of the brain, all subjects, and even all species. The purpose of this dissertation is 1) to apply ASL to a novel population to answer an important clinical question in the setting of Down syndrome, 2) to demonstrate proof of concept of a rapid technique to measure BBPC in mice to improve CBF quantification, and 3) to translate the correction method by applying it to a population of healthy canines using equipment and parameters suitable for use with humans. Chapter 2 reports the results of an ASL study of adults with Down syndrome (DS). This population is unique for their extremely high prevalence of Alzheimer’s disease (AD) and very low prevalence of systemic cardiovascular risk factors like atherosclerosis and hypertension. This prompted the hypothesis that AD pathology would lead to the development of perfusion deficits in people with DS despite their healthy cardiovascular profile. The results demonstrate that perfusion is not compromised in DS participants until the middle of the 6th decade of life after which measured global CBF was reduced by 31% (p=0.029). There was also significantly higher prevalence of residual arterial signal in older participants with DS (60%) than younger DS participants (7%, p = 0.005) or non-DS controls (0%, p < 0.001). This delayed pattern of perfusion deficits in people with DS differs from observations in studies of sporadic AD suggesting that adults with DS benefit from an improved cardiovascular risk profile early in life. Chapter 3 introduces calibrated short TR recovery (CaSTRR) imaging as a rapid method to measure BBPC and its development in mice. This was prompted by the inability to account for potential changes in BBPC due to age, brain atrophy, or the accumulation of hydrophobic A-β plaques in the ASL study of people with DS in Chapter 2. The CaSTRR method reduces acquisition time of BBPC maps by 87% and measures a significantly higher BBPC in cortical gray matter (0.99±0.04 mL/g,) than white matter in the corpus callosum (0.93±0.05 mL/g, p=0.03). Furthermore, when CBF maps are corrected for BBPC, the contrast between gray and white matter regions of interest is improved by 14%. This demonstrates proof of concept for the CaSTRR technique. Chapter 4 describes the application of CaSTRR on healthy canines (age 5-8 years) using a 3T human MRI scanner. This represents a translation of the technique to a setting suitable for use with a human subject. Both CaSTRR and pCASL acquisitions were performed and further optimization brought the acquisition time of CaSTRR down to 4 minutes which is comparable to pCASL. Results again show higher BBPC in gray matter (0.83 ± 0.05 mL/g) than white matter (0.78 ± 0.04 mL/g, p = 0.007) with both values unaffected by age over the range studied. Also, gray matter CBF is negatively correlated with age (p = 0.003) and BBPC correction improved the contrast to noise ratio by 3.6% (95% confidence interval = 0.6 – 6.5%). In summary, the quantification of ASL can be improved using BBPC maps derived from the novel, rapid CaSTRR technique.

Liver cancer ranks the third most common cause of cancer related mortality worldwide. Radiomicrosphere therapy (RMT), a form of radiation therapy, involves administration of Yttrium-90 (90Y) microspheres to the liver via the hepatic artery. 90Y microspheres bremsstrahlung SPECT/CT or PET/CT imaging could potentially identify an extrahepatic uptake. An early detection of such an uptake, thus, could initiate preventative measures early on. However, the quantitative accuracy of bremsstrahlung SPECT/CT images is limited by the wide and continuous energy spectrum of 90Y bremsstrahlung photons. 90Y PET/CT imaging is also possible but limited by the extremely small internal pair production decay. These limitation lead to inaccurate quantitation of microsphere biodistribution especially in small tumors. SPECT/CT and PET/CT acquisition of a Jasczak phantom with eight spherical inserts filled with 90Y3Cl solution were performed to measure the quantitative accuracy of the two imaging modalities. 90Y microsphere SPECT/CT data of 17 patients who underwent RMT for primary or metastatic liver cancer were acquired. Technetium-99m macroaggregated albumin (99mTc-MAA) SPECT/CT scans were also collected, but available for only twelve of the patients. SPECT/CT images from phantoms were used to determine the optimal iteration number for the iterative spatial resolution recovery algorithm. Methods for image based calculation of calibration factors for activity estimation from the patient and phantom 90Y bremsstrahlung SPECT/CT images were developed. Tumor areas were segmented using an active contour method. The 99mTc-MAA and 90Y microsphere SPECT/CT images were co-registered a priori for correlation analysis. Comparison of uptake on 99mTc-MAA and 90Y microsphere SPECT/CT images was assessed using tumor to healthy liver ratios. Furthermore, a three dimensional absorbed dose estimation algorithm was developed using the voxel S-value method. Absorbed doses within the tumor and healthy part of the liver were investigated for correlation with administered activity. Improvement in contrast to noise ratio and contrast recovery coefficients (QH) on patient and phantom 90Y bremsstrahlung SPECT/CT images as well as PET/CT images were achieved. Total activity estimations in liver and phantom gave mean percent errors of -4 ± 12% and -23 ± 41% for patient and phantom SPECT/CT studies. The pre and post-treatment images showed significant correlation (r = 0.9, p < 0.05) with mean TLR of 9.2 ± 9.4 and 5.0 ± 2.2 on 99mTc-MAA and 90Y microspheres SPECT/CT respectively. The correlation between the administered activity and tumor absorbed dose was weak (r = 0.5, p > 0.05), however, healthy liver absorbed dose increased with administered activity (r = 0.8, p < 0.05).

Ongoing developments in the field of medical imaging modalities have pushed the frontiers of modern medicine and biomedical engineering, prompting the need for new applications to improve diagnosis, treatment and prevention of diseases. Biomedical data visualization and modeling rely predominately on manual processing and utilization of voxel and facet based homogeneous models. Biological structures are naturally heterogeneous and in order to accurately design and biomimic biological structures, properties such as chemical composition, size and shape of biological constituents need to be incorporated in the computational biological models. Our proposed approach involves generating a density point cloud based on the intensity variations in a medical image slice, to capture tissue density variations through point cloud densities. The density point cloud is ordered and approximated with a set of cross-sectional least-squares B-Spline curves, based on which a skinned B-Spline surface is generated. The aim of this method is to capture and accurately represent density variations within the medical image data with a lofted surface function. The fitted B-Spline surface is sampled at uniformly distributed parameters, and our preliminary results indicate that the bio-CAD model preserves the density variations of the original image based point cloud. The resultant surface can thus be visualized by mapping the density in the parametric domain into color in pixel domain. The B-Spline function produced from each image slice can be used for medical visualization and heterogeneous tissue modeling. The process can be repeated for each slice in the medical dataset to produce heterogeneous B-Spline volumes. The emphasis of this research is placed on accuracy and shape fidelity needed for medical operations.

Radioactive seed implant brachytherapy is a common radiotherapy treatment method for prostate cancer. In current clinical practice, a seed consists of a single isotope, such as 125I or 103Pd. A seed containing a mixture of two isotopes has been proposed for prostate cancer treatment. This study investigates a method for defining a prescription dose for new seed compositions based on matching the biological equivalent dose (BED) of a reference plan. Ten prostate cancer cases previously treated using single isotope seeds (5 using 125I seeds and 5 using 103Pd seeds) were selected for this study. Verification of the method was done by calculating prescription doses for 103Pd and 125I seeds. A prescription dose for a 50/50 hybrid seed was calculated. Number and location of seeds remained invariant within each case. The BED distributions for hybrid and single isotope seed plans were generated and matched to the BED distribution generated off of the optimized plans. For the 125I isotopes, the dose necessary to cover 90% of the prostate with a BED of 110 Gy is 145 Gy. For the same BED coverage, the dose for 103Pd and 50/50 hybrid seed is 120 Gy and 137 Gy respectively. A method is introduced for obtaining prescription doses for new brachytherapy sources. The method was verified by obtaining doses for 125I and 103Pd isotopes which match clinical prescription doses. The method developed is robust enough to calculate prescription doses in any region of interest, for any seed type, and for any isotope as long as the BED coverage remains invariant with respect to the treatment plan. Numerical calculations were performed to derive analytical conversions of total dose to BED for 50/50, 75/25 and 25/75 hybrid seeds. These analytical conversions are faster than the original numerical methods employed allowing for real-time BED optimization for hybrid seeds. Varying seed distribution was seen not to influence the analytical conversions. It was observed that when total dose remained invariant while individual isotope contributions varied, the value of BED varied. The BED variance was seen to be the smaller at larger BED values (~2% at 100 Gy). Using the conversions derived in this paper, BED based optimization for hybrid seeds are now performable. However, these conversions should only be used in high dose regions due to high uncertainty in the low regime.