Medical imaging is a powerful tool for clinical practice allowing in-vivo insight into a patient’s disease state. Many modalities exist, allowing for the collection of diverse information about the underlying tissue structure and/or function. Traditionally, medical professionals use visual assessment of scans to search for disease, assess relevant disease predictors and propose clinical intervention steps. However, the imaging data contain potentially useful information beyond visual assessment by trained professional. To better use the full depth of information contained in the image sets, quantitative imaging characteristics (QICs), can be extracted using mathematical and statistical operations on regions or volumes of interests. The process of using QICs is a pipeline typically involving image acquisition, segmentation, feature extraction, set qualification and analysis of informatics. These descriptors can be integrated into classification methods focused on differentiating between disease states. Lung cancer, a leading cause of death worldwide, is a clear application for advanced in-vivo imaging based classification methods. We hypothesize that QICs extracted from spatially-linked and size-standardized regions of surrounding lung tissue can improve risk assessment quality over features extracted from only the lung tumor, or nodule, regions. We require a robust and flexible pipeline for the extraction and selection of disease QICs in computed tomography (CT). This includes creating an optimized method for feature extraction, reduction, selection, and predictive analysis which could be applied to a multitude of disease imaging problems. This thesis expanded a developmental pipeline for machine learning using a large multicenter controlled CT dataset of lung nodules to extract CT QICs from the nodule, surrounding parenchyma, and greater lung volume and explore CT feature interconnectivity. Furthermore, it created a validated pipeline that is more computationally and time efficient and with stability of performance. The modularity of the optimized pipeline facilitates broader application of the tool for applications beyond CT identified pulmonary nodules. We have developed a flexible and robust pipeline for the extraction and selection of Quantitative Imaging Characteristics for Risk Assessment from the Tumor and its Environment (QIC-RATE). The results presented in this thesis support our hypothesis, showing that classification of lung and breast tumors is improved through inclusion of peritumoral signal. Optimal performance in the lung application achieved with the QIC-RATE tool incorporating 75% of the nodule diameter equivalent in perinodular parenchyma with a development performance of 100% accuracy. The stability of performance was reflected in the maintained high accuracy (98%) in the independent validation dataset of 100 CT from a separate institution. In the breast QIC-RATE application, optimal performance was achieved using 25% of the tumor diameter in breast tissue with 90% accuracy in development, 82% in validation. We address the need for more complex assessments of medically imaged tumors through the QIC-RATE pipeline; a modular, scalable, transferrable pipeline for extracting, reducing and selecting, and training a classification tool based on QICs. Altogether, this research has resulted in a risk assessment methodology that is validated, stable, high performing, adaptable, and transparent.

Introduction: The use of CBCT to visualize the relationship between the inferior alveolar canal and the mandibular third molar roots continues to grow as it is becoming the standard of care. It becomes important to understand the impact that luminance, one of the factors that affects the viewing conditions of digital images, has on appropriately assessing the third molar-canal relationship. To date, no study has assessed the impact of luminance on visualizing anatomic structures on CBCT. The aim of this study is to determine if there is a difference in the ability to appropriately assess the root development and the third molar-canal relationship on a medical grade monitor with four different luminance settings on CBCT. Materials and methods: 285 scans were randomized and evaluated by three calibrated and masked evaluators. The evaluations were completed on a Barco MDNC-3321 Nio Color 3MP monitor (Kortrijk, Belgium) monitor at four different luminance settings; 200 cd/m2, 300 cd/m2, 400 cd/m2, and 500 cd/m2. The gold standard was established by two board-certified oral and maxillofacial radiologists. All evaluations were performed in a controlled subdued environment lighting of less than 15 lux. There was a washout period of at least one week between each of the four evaluations by an observer. Results: The accuracy of two of the three evaluators was substantial to almost perfect independent of luminance. None of these assessments showed any statistical significance (P = 0.05). The accuracy of one evaluator was moderate to almost perfect for all evaluations with one assessment of one canal showing statistical significance (P = 0.05). Conclusion: There is no difference in the ability to appropriately assess the third molar canal relationship or the root development of third molars on a medical grade monitor at luminance settings between the range of 200 cd/m2 and 500 cd/m2 when viewed in a dimly lit room.

With the introduction of four dimensional-computed tomography (4DCT), treatment centers are now better able to account for respiration-induced uncertainty in radiation therapy treatment planning for lung cancer. We examined two practices in which 4DCT is used in radiotherapy. Our first study investigated the dosimetric uncertainty in four-dimensional (4D) dose calculation using three temporal probability distributions: 1) uniform distribution, 2) sinusoidal distribution, and 3) patient-specific distribution derived from the respiratory trace. Four-dimensional dose was evaluated in nine lung cancer patients. First, dose was computed for each of 10 binned CTs using 4DCT and deformable image registration. Next, the 10 deformed doses were summed together using one of three temporal probability distributions. To compare the two approximated 4D dose calculations to the 4D calculation derived using the patient's respiratory trace, 3D gamma analysis was performed using a tolerance criteria of 3% dose difference and 3mm distance to agreement. Additionally, mean lung dose (MLD), mean tumor dose (MTD), and lung V20 were used to assess clinical impact. For all patients, both uniform and sinusoidal dose distributions were found to have an average gamma passing rate >99% for both the lung and PTV volumes. Compared with 4D dose calculated using the patient respiratory trace, uniform distribution and sinusoidal distribution showed a percentage difference on average of -0.1±0.6% and -0.2±0.4% in MTD, -0.2±2.0% and -0.2±1.3% in MLD, 0.9±2.8% and -0.7±1.8% in lung V20, respectively. We concluded that 4D dose computed using either a uniform or sinusoidal temporal probability distribution is able to approximate 4D dose computed using the patient-specific respiratory trace. Our second study evaluated the dosimetric and temporal effects of respiratory gated radiation therapy using four different gating windows (20EX-20IN, 40EX-40IN, 60EX-60IN, and 80EX-80IN) and estimated the corresponding treatment delivery times for normal (500MU/min) and high (1500MU/min) dose rates. Five patients (3 non-gated, 2 gated 80EX-80IN) were retrospectively evaluated. For each patient, four individual treatment plans corresponding to the four different gating windows were created, and treatment delivery time for each plan was estimated using a MATLAB (MathWorks, Natick, MA) algorithm. Results showed that smaller gating windows reduced PTV volume, mean lung dose, and lung V20, while maintaining mean tumor dose and PTV coverage. Treatment times for gated plans were longer when dose rate was unchanged, however, increased dose rates were shown to achieve treatment times comparable to or faster than non-gated delivery times. We concluded that gated radiation therapy in lung cancer patients could potentially reduce lung toxicity, while as effectively treating the target volume. Furthermore, increased dose rates with gated radiation therapy are able to provide treatment times comparable to non-gated treatment.

CT is a powerful method for noninvasive assessment of the lung. Advancements to CT technology have guided the high-resolution structural and functional assessment of lung diseases. This has helped make the transition from characterizing the severity of lung disease to novel phenotyping of disease subtypes. Chronic obstructive pulmonary disease (COPD) is a spectrum of inflammatory lung disease affecting lung parenchyma, airways, and the pulmonary and systemic vasculature. Quantitative CT-based measures have largely focused on quantifying the extent of airway and parenchymal damage with disease. Recently perfusion CT method has been used to assess the pulmonary vascular bed. This technique was used to demonstrate a vascular etiology of smoking-associated centriacinar emphysema (CAE), a subtype of the COPD spectrum. However, technical challenges have limited the transition of this CT method to clinical studies to assess pulmonary vascular physiology. In this thesis, we introduce dual energy CT-perfused blood volume (DECT-PBV) as a novel image-based biomarker to assess peripheral pulmonary vascular dysfunction. Using this technique, we show that smoking-associated pulmonary perfusion heterogeneity, a marker of abnormal blood flow is a reversible process, in the midst of smoking-associated lung inflammation, and not a product of advanced lung disease. We demonstrate, via regional PBV measures and structural measures of the central pulmonary vessels, that the reversibility of pulmonary perfusion heterogeneity is a direct result of increased peripheral (downstream) parenchymal perfusion. We validate our quantitative imaging approach in a unique cohort of early CAE-susceptible smokers using a pharmaceutical intervention to dilate the pulmonary parenchymal vascular bed. The validated DECT approach and our novel DECT imaging findings extend our characterization of the vascular phenotype in inflammatory lung disease and provide a framework for future quantitative imaging studies of the lung to assess early intervention targeted to pulmonary vessels.

Purpose: The purpose of this study was to assess the radiographic changes in periapical status and analysis of healing determined using periapical radiographs (PA) versus cone beam computed tomography (CBCT) pre-operatively and at 3-64 months following endodontic treatment. Methods: Pre/post treatment radiograph and CBCT scans of patients who had NSRCT, NSReTx, or SRCT from July 2011-December 2018 at VCU Graduate Endodontic clinic were included in this study. Volumetric and linear measurements of periapical lesions on initial and recall PA and CBCT images were performed using three calibrated examiners. Changes and differences in the estimated area from PA to CBCT were compared using the Wilcoxon signed-rank test. McNemar’s chi-squared test was used to determine agreement in the proportion of lesions that were absent (0x0) between the PA and corresponding view of CBCT. This data was used to calculate the sensitivity, specificity, positive predictive value (PPV), and negative predicative value (NPV). Results: A total of 51 patients with a median healing time of 13 months were included in the analysis. Significant healing was observed on both PA and CBCT images (p-value Conclusion: Assessment using CBCT revealed a lower healing rate for all treatment categories compared with periapical radiographs. CBCT was more likely to detect the presence of a PARL, whereas a periapical radiograph would be less sensitive to detection of a PARL. Significant healing cannot be detected at an earlier point in time with PA radiographs or CBCT.

Objectives: The aim of the current study is to establish the reliability of linear cephalometric measurements made on mandibles and their respective 3D printed models created from different voxel resolutions from a cone beam CT machine. Materials and methods: Ten dry mandibles obtained from the Department of Oral Pathology, Radiology and Medicine at The University of Iowa College Of Dentistry were used for this study. All mandibles were scanned on the i-CAT FLX cone beam CT machine (Imaging Sciences International, LLC, Pennsylvania, USA) using voxel resolutions of .30mm, .25mm and .20 mm in a 16cm x 8cm field of view using 360° rotation. The 3D models were reconstructed and saved as .STL files using 3D Slicer software and send to a 3D printer for printing. Two observers measured the 10 mandibles and 30 3D printed models. The measurement were repeated on 50% of the samples after at least one week interval. Cronbach’s alpha and intraclass correlation coefficient were calculated to measure reliability. Results: Good to excellent interobserver and intraobserver reliability was achieved across most of the measurements. There was no difference in reliability across models made from different voxel sizes. Conclusion: The current study successfully showed that the reliability of measurements made on 3D printed models of dry skull mandibles created using fused deposition modeling technique using images of different voxel sizes from an i-CAT FLX CBCT machine are valid, reproducible, and reliable and can be used for diagnostic and clinical purposes.

Currently in our clinic, a mega-voltage cone beam computed tomography (MVCBCT) scan is performed before each treatment for patient localization. For non-small cell lung cancer (NSCLC) patients, a strain gauge is used as an external surrogate to indicate tumor motion in both the planning stage and the treatment stage. However, it is likely that the amplitude of tumor motion varies between treatment fractions without a corresponding change in the surrogate signal. Motion amplitude larger than what was planned may underdose the tumor and overexpose normal tissues. The overall objective of this project is to extend the capabilities of MVCBCT for respiratory motion management by taking advantage of 2D projection images. First, a new method was developed to detect ipsi-lateral hemi-diaphragm apex (IHDA) motion along superior-inferior (SI) direction in 3D. Then a respiratory correlated reconstruction method was implemented and verified. This method is able to create MVCBCT volume in the full exhale (FE) and the full inhale (FI) phases, respectively. The diaphragm to tumor motion ratio (DTMR) was derived by quantifying the absolute position of the tumor and IHDA in these two volumes. The DTMR and the extracted IHDA motion were further used to calibrate the strain gauge signal. Second, an organ motion detection approach was developed, in which the detection is converted into an optimal interrelated surface detection problem. The framework was first applied to tumor motion extraction, which enables accurate detection for large tumors (with a diameter not smaller than 1.9cm). The framework was then applied to lung motion extraction and the extracted lung motion model was used to create a series of displacement vector fields for a motion compensated (MC) reconstruction. The accuracy of both tumor extraction and the MC approach was validated, which shows their clinical feasibility. Last but not least, a novel enhancement framework was developed. The aim of this approach is to eliminate the overlapping tissues and organs in the CBCT projection images. Though scattering and noise is the major problem, the proposed method is able to achieve enhanced projection images with a higher contrast to noise ratio (CNR) without compromising detection accuracy on tumors and IHDA.

X-ray computed tomography is a widely adopted medical imaging method that uses projections to recover the internal image of a subject. Since the invention of X-ray computed tomography in the 1970s, several generations of CT scanners have been developed. As 3D-image reconstruction increases in popularity, the long processing time associated with these machines has to be significantly reduced before they can be practically employed in everyday applications. Parallel computing is a computer science computing technique that utilizes multiple computer resources to process a computational task simultaneously; each resource computes only a part of the whole task thereby greatly reducing computation time. In this thesis, we use parallel computing technology to speed up the reconstruction while preserving the image quality. Three representative reconstruction algorithms--namely, Katsevich, EM, and Feldkamp algorithms--are investigated in this work. With the Katsevich algorithm, a distributed-memory PC cluster is used to conduct the experiment. This parallel algorithm partitions and distributes the projection data to different computer nodes to perform the computation. Upon completion of each sub-task, the results are collected by the master computer to produce the final image. This parallel algorithm uses the same reconstruction formula as the sequential counterpart, which gives an identical image result. The parallelism of the iterative CT algorithm uses the same PC cluster as in the first one. However, because it is based on a local CT reconstruction algorithm, which is different from the sequential EM algorithm, the image results are different with the sequential counterpart. Moreover, a special strategy using inhomogeneous resolution was used to further speed up the computation. The results showed that the image quality was largely preserved while the computational time was greatly reduced. Unlike the two previous approaches, the third type of parallel implementation uses a shared-memory computer. Three major accelerating methods--SIMD (Single instruction, multiple data), multi-threading, and OS (ordered subsets)--were employed to speed up the computation. Initial investigations showed that the image quality was comparable to those of the conventional approach though the computation speed was significantly increased.

Cancer is the second deadliest disease in the United States with an estimated 1.69 million new cases in 2017. Medical imaging modalities, such as computed tomography (CT) and magnetic resonance imaging (MRI), are widely used in clinical medicine to detect, diagnose, plan treatment, and monitor tumors within the body. Advances in imaging research related to cancer assessment have largely relied on consented human patients, often including varied populations and treatments. Tumor bearing mouse models have been highly valued for basic science research, but imaging focused applications are limited by the translational ability of micro imaging systems. Pig models are well suited to bridge the gap between human cohorts and mouse models due to similar anatomy, physiology, life-span, and size between pigs and humans. These models provide the opportunity to advance medical imaging while simultaneously characterizing progressive changes resulting from an intervention, exposure, or genetic modification. We present a foundation for effectively characterizing disease models in pigs, susceptible to tumor development, using longitudinal medical image acquisition and post-processing techniques for quantification of disease. Longitudinal, whole-body protocols were developed with CT and MRI. Focus was placed on systematic process, including transportation, anesthesia and positioning, imaging, and environmental controls. Demonstration of the methodology was achieved with six pigs (30-85 kg) with four to seven imaging time points acquired per animal. Consistent positioning across time points (CT to CT) and within time points (CT to MRI) was assessed with distance measures obtained from the skeleton following rigid registration between images. Alignment across time points was achieved with an average value of 16.51 (± 12.46) mm observed all acquired measurements. For consistent, retrievable, and complete qualitative assessment of acquired images, structured reports were developed, including assessment of imaging quality and emphasis on tumor development throughout the body. Reports were used to perform a systematic, semi-qualitative comparison of CT and MRI lung assessment with an overall agreement of 72% in detection of disease indicators. A multi-level registration algorithm was developed to align anatomic structures of interest in the acquired longitudinal datasets. The algorithm consisted of initialization followed by repeated application of a core registration framework as the input data reduced in image field of view. It was applied to align regions of interest in the brain, upper right lung, and right kidney. Validation was performed with overlap (range = [0.0,1.0], complete overlap = 1) and distance measures (range = [0.0, ∞], perfect match = 0.0) of corresponding segmentations with overall results of 0.85 (± 0.11) and 0.41 (± 0.83) mm, respectively. An extension of the algorithm was created, demonstrating the ability to incorporate directional growth and feature extraction measurements into longitudinal tumor progression monitoring. Techniques were applied to a phantom dataset showing solid tumor growth and transition from a non-solid to part-solid lesion in the lungs. Finally, the developed methods – imaging, structured reporting, registration, and longitudinal feature extraction – were applied to four different porcine models pre-disposed to tumor development. 1) A genetically modified Li-Fraumeni (TP53R167H/+/TP53R167H/R167H) background model showing the development of osteosarcoma and lymphoma. 2) A TP53R167H/+ animal with exposure to crystalline silica showing progression of silicosis in the lungs. 3) TP53R167H/+/TP53R167H/R167H animals with exposure to radiation for targeted sarcoma development and 4) TP53R167H/+ pigs with conditional KRASG12D/+ mutation activated in the lung and pancreas. Whole-body and targeted imaging protocols were developed for each model and qualitatively interpreted by a radiologist using structured reports. Multi-level registration was used to align identified tumors and longitudinal features were extracted to quantitatively track change over time. Overall, the developed methods aided in the effective, non-invasive characterization of these animals.

The aim of this study is to investigate the functional and structural differences between normal subjects and asthmatics via image registration and computational fluid dynamics (CFD), together with pulmonary function test's (PFT) and one-image-based variables. We analyzed three populations of CT images: 50 normal, 42 non-severe asthmatic and 52 severe asthmatic subjects at total lung capacity (TLC) and functional residual capacity (FRC). A mass preserving image registration technique was employed to match CT images at TLC and FRC for assessments of regional volume change and anisotropic deformation. Instead of existing threshold-based air-trapping measure, a fraction-based air-trapping measure was proposed to account for inter-site and inter-subject variations of CT density. We also analyzed structural alterations of asthmatic airways, including bifurcation angle, hydraulic diameter, luminal area and wall area. CFD and particle tracking simulations are employed with physiologically-consistent boundary condition. As compared with normal subjects, severe asthmatics exhibit reduced air volume change (consistent with air-trapping) and more isotropic deformation in the basal lung regions, but increased air volume change associated with increased anisotropic deformation in the apical lung regions. In the multi-center study, the traditional air-trapping measure showed the significant site-variability due to the differences of scanners and coaching methods. The proposed fraction-based air-trapping measure is able to overcome the inter-site and inter-subject variations, allowing analysis of large data sets collected from multiple centers. We further demonstrate alterations of bifurcation angle, constriction, wall thickness and non-circularity at local branch level in severe asthmatics. The bifurcation angle, non-circularity and especially reduced hydraulic diameter significantly affect the increase of particle deposition in severe asthmatics. In summary, the two-image registration-based deformation provides a tool for distinguishing differences in lung mechanics among populations. The new fraction-based air-trapping measure significantly improves the association of air-trapping with the presence and severity of asthma and the correlation with forced expiratory volume in 1 second over forced vital capacity (FEV1/FVC) than existing approaches. The altered functions and structures such as air-volume change, branching angles, non-circular shapes, wall thickness and hydraulic diameters that found in asthmatics are strongly associated with the flow structures and particle depositions.