Repeatability of computed tomography liver radiomic features in a nonhuman primate model of diet-induced steatosis.

Purpose: We describe a method to identify repeatable liver computed tomography (CT) radiomic features, suitable for detection of steatosis, in nonhuman primates. Criteria used for feature selection exclude nonrepeatable features and may be useful to improve the performance and robustness of radiomics-based predictive models. Approach: Six crab-eating macaques were equally assigned to two experimental groups, fed regular chow or an atherogenic diet. High-resolution CT images were acquired over several days for each macaque. First-order and second-order radiomic features were extracted from six regions in the liver parenchyma, either with or without liver-to-spleen intensity normalization from images reconstructed using either a standard (B-filter) or a bone-enhanced (D-filter) kernel. Intrasubject repeatability of each feature was assessed using a paired t-test for all scans and the minimum p-value was identified for each macaque. Repeatable features were defined as having a minimum p-value among all macaques above the significance level after Bonferroni's correction. Features showing a significant difference with respect to diet group were identified using a two-sample t-test. Results: A list of repeatable features was generated for each type of image. The largest number of repeatable features was achieved from spleen-normalized D-filtered images, which also produced the largest number of second-order radiomic features that were repeatable and different between diet groups. Conclusions: Repeatability depends on reconstruction kernel and normalization. Features were quantified and ranked based on their repeatability. Features to be excluded for more robust models were identified. Features that were repeatable but different between diet groups were also identified.

Novel machine-learning analysis of SARS-CoV-2 infection in a subclinical nonhuman primate model using radiomics and blood biomarkers.

Detection of the physiological response to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is challenging in the absence of overt clinical signs but remains necessary to understand a full subclinical disease spectrum. In this study, our objective was to use radiomics (from computed tomography images) and blood biomarkers to predict SARS-CoV-2 infection in a nonhuman primate model (NHP) with inapparent clinical disease. To accomplish this aim, we built machine-learning models to predict SARS-CoV-2 infection in a NHP model of subclinical disease using baseline-normalized radiomic and blood sample analyses data from SARS-CoV-2-exposed and control (mock-exposed) crab-eating macaques. We applied a novel adaptation of the minimum redundancy maximum relevance (mRMR) feature-selection technique, called mRMR-permute, for statistically-thresholded and unbiased feature selection. Through performance comparison of eight machine-learning models trained on 14 feature sets, we demonstrated that a logistic regression model trained on the mRMR-permute feature set can predict SARS-CoV-2 infection with very high accuracy. Eighty-nine percent of mRMR-permute selected features had strong and significant class effects. Through this work, we identified a key set of radiomic and blood biomarkers that can be used to predict infection status even in the absence of clinical signs. Furthermore, we proposed and demonstrated the utility of a novel feature-selection technique called mRMR-permute. This work lays the foundation for the prediction and classification of SARS-CoV-2 disease severity.

Magnetic Resonance Imaging for Monitoring of Hepatic Disease Induced by Ebola Virus: a Nonhuman Primate Proof-of-Concept Study.

Severe liver impairment is a well-known hallmark of Ebola virus disease (EVD). However, the role of hepatic involvement in EVD progression is understudied. Medical imaging in established animal models of EVD (e.g., nonhuman primates [NHPs]) can be a strong complement to traditional assays to better investigate this pathophysiological process in vivo and noninvasively. In this proof-of-concept study, we used longitudinal multiparametric magnetic resonance imaging (MRI) to characterize liver morphology and function in nine rhesus monkeys after exposure to Ebola virus (EBOV). Starting 5 days postexposure, MRI assessments of liver appearance, morphology, and size were consistently compatible with the presence of hepatic edema, inflammation, and congestion, leading to significant hepatomegaly at necropsy. MRI performed after injection of a hepatobiliary contrast agent demonstrated decreased liver signal on the day of euthanasia, suggesting progressive hepatocellular dysfunction and hepatic secretory impairment associated with EBOV infection. Importantly, MRI-assessed deterioration of biliary function was acute and progressed faster than changes in serum bilirubin concentrations. These findings suggest that longitudinal quantitative in vivo imaging may be a useful addition to standard biological assays to gain additional knowledge about organ pathophysiology in animal models of EVD. IMPORTANCE Severe liver impairment is a well-known hallmark of Ebola virus disease (EVD), but the contribution of hepatic pathophysiology to EVD progression is not fully understood. Noninvasive medical imaging of liver structure and function in well-established animal models of disease may shed light on this important aspect of EVD. In this proof-of-concept study, we used longitudinal magnetic resonance imaging (MRI) to characterize liver abnormalities and dysfunction in rhesus monkeys exposed to Ebola virus. The results indicate that in vivo MRI may be used as a noninvasive readout of organ pathophysiology in EVD and may be used in future animal studies to further characterize organ-specific damage of this condition, in addition to standard biological assays.

Computed Tomography Imaging for Monitoring of Marburg Virus Disease: a Nonhuman Primate Proof-Of-Concept Study.

Marburg virus (MARV) is a highly virulent zoonotic filovirid that causes Marburg virus disease (MVD) in humans. The pathogenesis of MVD remains poorly understood, partially due to the low number of cases that can be studied, the absence of state-of-the-art medical equipment in areas where cases are reported, and limitations on the number of animals that can be safely used in experimental studies under maximum containment animal biosafety level 4 conditions. Medical imaging modalities, such as whole-body computed tomography (CT), may help to describe disease progression in vivo, potentially replacing ethically contentious and logistically challenging serial euthanasia studies. Towards this vision, we performed a pilot study, during which we acquired whole-body CT images of 6 rhesus monkeys before and 7 to 9 days after intramuscular MARV exposure. We identified imaging abnormalities in the liver, spleen, and axillary lymph nodes that corresponded to clinical, virological, and gross pathological hallmarks of MVD in this animal model. Quantitative image analysis indicated hepatomegaly with a significant reduction in organ density (indicating fatty infiltration of the liver), splenomegaly, and edema that corresponded with gross pathological and histopathological findings. Our results indicated that CT imaging could be used to verify and quantify typical MVD pathogenesis versus altered, diminished, or absent disease severity or progression in the presence of candidate medical countermeasures, thus possibly reducing the number of animals needed and eliminating serial euthanasia. IMPORTANCE Marburg virus (MARV) is a highly virulent zoonotic filovirid that causes Marburg virus disease (MVD) in humans. Much is unknown about disease progression and, thus, prevention and treatment options are limited. Medical imaging modalities, such as whole-body computed tomography (CT), have the potential to improve understanding of MVD pathogenesis. Our study used CT to identify abnormalities in the liver, spleen, and axillary lymph nodes that corresponded to known clinical signs of MVD in this animal model. Our results indicated that CT imaging and analyses could be used to elucidate pathogenesis and possibly assess the efficacy of candidate treatments.

Toward the determination of sensitive and reliable whole-lung computed tomography features for robust standard radiomics and delta-radiomics analysis in a nonhuman primate model of coronavirus disease 2019.

Purpose: We propose a method to identify sensitive and reliable whole-lung radiomic features from computed tomography (CT) images in a nonhuman primate model of coronavirus disease 2019 (COVID-19). Criteria used for feature selection in this method may improve the performance and robustness of predictive models. Approach: Fourteen crab-eating macaques were assigned to two experimental groups and exposed to either severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) or a mock inoculum. High-resolution CT scans were acquired before exposure and on several post-exposure days. Lung volumes were segmented using a deep-learning methodology, and radiomic features were extracted from the original image. The reliability of each feature was assessed by the intraclass correlation coefficient (ICC) using the mock-exposed group data. The sensitivity of each feature was assessed using the virus-exposed group data by defining a factor R that estimates the excess of variation above the maximum normal variation computed in the mock-exposed group. R and ICC were used to rank features and identify non-sensitive and unstable features. Results: Out of 111 radiomic features, 43% had excellent reliability ( ICC > 0.90 ), and 55% had either good ( ICC > 0.75 ) or moderate ( ICC > 0.50 ) reliability. Nineteen features were not sensitive to the radiological manifestations of SARS-CoV-2 exposure. The sensitivity of features showed patterns that suggested a correlation with the radiological manifestations. Conclusions: Features were quantified and ranked based on their sensitivity and reliability. Features to be excluded to create more robust models were identified. Applicability to similar viral pneumonia studies is also possible.

Atlas-based liver segmentation for nonhuman primate research.

PURPOSE: Certain viral infectious diseases cause systemic damage and the liver is an important organ affected directly by the virus and/or the hosts' response to the virus. Medical imaging indicates that the liver damage is heterogenous, and therefore, quantification of these changes requires analysis of the entire organ. Delineating the liver in preclinical imaging studies is a time-consuming and difficult task that would benefit from automated liver segmentation. METHODS: A nonhuman primate atlas-based liver segmentation method was developed to support quantitative image analysis of preclinical research. A set of 82 computed tomography (CT) scans of nonhuman primates with associated manual contours delineating the liver was generated from normal and abnormal livers. The proposed technique uses rigid and deformable registrations, a majority vote algorithm, and image post-processing operations to automate the liver segmentation process. This technique was evaluated using Dice similarity, Hausdorff distance measures, and Bland-Altman plots. RESULTS: Automated segmentation results compare favorably with manual contouring, achieving a median Dice score of 0.91. Limits of agreement from Bland-Altman plots indicate that liver changes of 3 Hounsfield units (CT) and 0.4 SUVmean (positron emission tomography) are detectable using our automated method of segmentation, which are substantially less than changes observed in the host response to these viral infectious diseases. CONCLUSION: The proposed atlas-based liver segmentation technique is generalizable to various sizes and species of nonhuman primates and facilitates preclinical infectious disease research studies. While the image analysis software used is commercially available and facilities with funding can access the software to perform similar nonhuman primate liver quantitative analyses, the approach can be implemented in open-source frameworks as there is nothing proprietary about these methods.

Meningeal blood-brain barrier disruption in acute traumatic brain injury.

The meninges serve as a functional barrier surrounding the brain, critical to the immune response, and can be compromised following head trauma. Meningeal enhancement can be detected on contrast-enhanced MRI in patients presenting with acute traumatic brain injury, even when head CT is negative. Following head trauma, gadolinium-based contrast appears to extravasate from the vasculature, enhancing the dura within minutes, and later permeates the subarachnoid space. The aims of this study were to characterize the initial kinetics of the uptake of contrast agent after injury and the delayed redistribution of contrast enhancement in the subarachnoid space in hyperacute patients. Neuroimaging was obtained prospectively in two large ongoing observational studies of patients aged 18 years or older presenting to the emergency department with suspected acute head injury. Dynamic contrast-enhanced MRI studies in a cohort of consecutively enrolling patients with mild traumatic brain injury (n = 36) determined that the kinetic half-life of dural-related meningeal enhancement was 1.3 ± 0.6 min (95% enhancement within 6 min). The extravasation of contrast into the subarachnoid space was investigated in a cohort of CT negative mild traumatic brain injury patients initially imaged within 6 h of injury (hyperacute) who subsequently underwent a delayed MRI, with no additional contrast administration, several hours after the initial MRI. Of the 32 patients with delayed post-contrast imaging, 18 (56%) had conspicuous expansion of the contrast enhancement into the subarachnoid space, predominantly along the falx and superior sagittal sinus. Patients negative for traumatic meningeal enhancement on initial hyperacute MRI continued to have no evidence of meningeal enhancement on the delayed MRI. These studies demonstrate that (i) the initial enhancement of the traumatically injured meninges occurs within minutes of contrast injection, suggesting highly permeable meningeal vasculature, and that (ii) contrast in the meninges redistributes within the subarachnoid space over the period of hours, suggesting a compromise in the blood-brain and/or blood-cerebrospinal barriers. Data from the parent study indicate that up to one in two patients with mild traumatic brain injury have traumatic brain injury on acute (<48 h) MRI, with a higher prevalence seen in patients with moderate or severe traumatic brain injury. The current study's findings of traumatic meningeal enhancement and the subsequent delayed extravasation of contrast into the subarachnoid spaces indicate that a substantial percentage of patients with even mild traumatic brain injury may have a transient disruption in barriers separating the vasculature from the brain.

The Use of Large-Particle Aerosol Exposure to Nipah Virus to Mimic Human Neurological Disease Manifestations in the African Green Monkey.

Nipah virus (NiV) is an emerging virus associated with outbreaks of acute respiratory disease and encephalitis. To develop a neurological model for NiV infection, we exposed 6 adult African green monkeys to a large-particle (approximately 12 µm) aerosol containing NiV (Malaysian isolate). Brain magnetic resonance images were obtained at baseline, every 3 days after exposure for 2 weeks, and then weekly until week 8 after exposure. Four of six animals showed abnormalities reminiscent of human disease in brain magnetic resonance images. Abnormalities ranged from cytotoxic edema to vasogenic edema. The majority of lesions were small infarcts, and a few showed inflammatory or encephalitic changes. Resolution or decreased size in some lesions resembled findings reported in patients with NiV infection. Histological lesions in the brain included multifocal areas of encephalomalacia, corresponding to known ischemic foci. In other regions of the brain there was evidence of vasculitis, with perivascular infiltrates of inflammatory cells and rare intravascular fibrin thrombi. This animal model will help us better understand the acute neurological features of NiV infection and develop therapeutic approaches for managing disease caused by NiV infection.

Unsteady wall shear stress analysis from image-based computational fluid dynamic aneurysm models under Newtonian and Casson rheological models.

The aim of this work was to determine whether or not Newtonian rheology assumption in image-based patient-specific computational fluid dynamics (CFD) cerebrovascular models harboring cerebral aneurysms may affect the hemodynamics characteristics, which have been previously associated with aneurysm progression and rupture. Ten patients with cerebral aneurysms with lobulations were considered. CFD models were reconstructed from 3DRA and 4DCTA images by means of region growing, deformable models, and an advancing front technique. Patient-specific FEM blood flow simulations were performed under Newtonian and Casson rheological models. Wall shear stress (WSS) maps were created and distributions were compared at the end diastole. Regions of lower WSS (lobulation) and higher WSS (neck) were identified. WSS changes in time were analyzed. Maximum, minimum and time-averaged values were calculated and statistically compared. WSS characterization remained unchanged. At high WSS regions, Casson rheology systematically produced higher WSS minimum, maximum and time-averaged values. However, those differences were not statistically significant. At low WSS regions, when averaging over all cases, the Casson model produced higher stresses, although in some cases the Newtonian model did. However, those differences were not significant either. There is no evidence that Newtonian model overestimates WSS. Differences are not statistically significant.

Understanding the role of hemodynamics in the initiation, progression, rupture, and treatment outcome of cerebral aneurysm from medical image-based computational studies.

About a decade ago, the first image-based computational hemodynamic studies of cerebral aneurysms were presented. Their potential for clinical applications was the result of a right combination of medical image processing, vascular reconstruction, and grid generation techniques used to reconstruct personalized domains for computational fluid and solid dynamics solvers and data analysis and visualization techniques. A considerable number of studies have captivated the attention of clinicians, neurosurgeons, and neuroradiologists, who realized the ability of those tools to help in understanding the role played by hemodynamics in the natural history and management of intracranial aneurysms. This paper intends to summarize the most relevant results in the field reported during the last years.

Template-based B1 inhomogeneity correction in 3T MRI brain studies.

Low noise, high resolution, fast and accurate T1 maps from MRI images of the brain can be performed using a dual flip angle method. However, B1 field inhomogeneity, which is particularly problematic at high field strengths (e.g., 3T), limits the ability of the scanner to deliver the prescribed flip angle, introducing errors into the T1 maps that limit the accuracy of quantitative analyses based on those maps. A dual repetition time method was used for acquiring a B1 map to correct that inhomogeneity. Additional inaccuracies due to misregistration of the acquired T1-weighted images were corrected by rigid registration, and the effects of misalignment on the T1 maps were compared to those of B1 inhomogeneity in 19 normal subjects. However, since B1 map acquisition takes up precious scanning time and most retrospective studies do not have B1 map, we designed a template-based correction strategy. B1 maps from different subjects were aligned using a twelve-parameter affine registration. Recomputed T1 maps showed an important improvement with respect to the noncorrected maps: histograms of all corrected maps exhibited two peaks corresponding to white and gray matter tissues, while unimodal histograms were observed in all uncorrected maps because of the inhomogeneity. A method to detect the best nonsubject-specific B1 correction based on a set of features was designed. The optimum set of weighting factors for those features was computed. The best available B1 correction was detected in almost all subjects while corrections comparable to the T1 map corrected using the B1 map from the same subject were detected in the others.

Patient-specific computational modeling of cerebral aneurysms with multiple avenues of flow from 3D rotational angiography images.

RATIONALE AND OBJECTIVES: Previous studies of aneurysm flow dynamics based on three-dimensional (3D) rotational angiography (RA) images were limited to aneurysms with a single route of blood inflow. However, aneurysms of the circle of Willis frequently involve locations with more than one source of inflow, such as aneurysms of the anterior communicating artery. The highest resolution images of cerebral vessels are from RA images, but this technique is limited to visualizing only one route of inflow at a time, leaving a significant limitation in the application of 3DRA image sets for clinical studies of patient-specific computational fluid dynamics (CFD) simulations. In this report, subject-specific models of cerebral aneurysms with multiple avenues of flow are constructed from RA images by using a novel combination of image co-registration and surface merging techniques. MATERIALS AND METHODS: RA images are obtained by means of contrast injection in each vessel that provides inflow to the aneurysm. Anatomic models are constructed independently of each of these vascular trees and fused together into a single model. The model is used to construct a finite element grid for CFD simulations of hemodynamics. RESULTS: Three examples of patient-specific models are presented: an anterior communicating artery aneurysm, a basilar tip aneurysm, and a model of an entire circle of Willis with five coincident aneurysms. The method is evaluated with a numeric phantom of an aneurysm in the anterior communicating artery. CONCLUSION: These examples show that this new technique can be used to create merged network numeric models for CFD modeling. Furthermore, intra-aneurysmal flow patterns are influenced strongly by merging of the two inflow streams. This effect decreases as distance from the merging streams increases.

Characterization of cerebral aneurysms for assessing risk of rupture by using patient-specific computational hemodynamics models.

BACKGROUND AND PURPOSE: Hemodynamic factors are thought to play an important role in the initiation, growth, and rupture of cerebral aneurysms. This report describes a pilot clinical study of the association between intra-aneurysmal hemodynamic characteristics from computational fluid dynamic models and the rupture of cerebral aneurysms. METHODS: A total of 62 patient-specific models of cerebral aneurysms were constructed from 3D angiography images. Computational fluid dynamics simulations were performed under pulsatile flow conditions measured on a normal subject. The aneurysms were classified into different categories, depending on the complexity and stability of the flow pattern, the location and size of the flow impingement region, and the size of the inflow jet. The 62 models consisted of 25 ruptured and 34 unruptured aneurysms and 3 cases with unknown histories of hemorrhage. The hemodynamic features were analyzed for associations with history of rupture. RESULTS: A large variety of flow patterns was observed: 72% of ruptured aneurysms had complex or unstable flow patterns, 80% had small impingement regions, and 76% had small jet sizes. By contrast, unruptured aneurysms accounted for 73%, 82%, and 75% of aneurysms with simple stable flow patterns, large impingement regions, and large jet sizes, respectively. Aneurysms with small impingement sizes were 6.3 times more likely to have experienced rupture than those with large impingement sizes (P = .01). CONCLUSIONS: Image-based patient-specific numeric models can be constructed in an efficient manner that allows clinical studies of intra-aneurysmal hemodynamics. A simple flow characterization system was proposed, and interesting trends in the association between hemodynamic features and aneurysmal rupture were found. Simple stable patterns, large impingement regions, and jet sizes were more commonly seen with unruptured aneurysms. By contrast, ruptured aneurysms were more likely to have disturbed flow patterns, small impingement regions, and narrow jets.

Characterization of shear stress on the wall of the carotid artery using magnetic resonance imaging and computational fluid dynamics.

Considerable evidence has emerged that adverse blood flow patterns are a major factor in the onset of atherosclerotic disease and may play a role in disease progression. This chapter reviews a technique, referred to as vascular computational fluid dynamics (CFD), for characterizing blood flow patterns in large arteries from magnetic resonance angiography (MRA) and velocity-encoded phase-contrast magnetic resonance (PC MR) imaging. In vascular CFD, hemodynamic conditions are modeled by the finite-element method with flow is governed by the incompressible Navier-Stokes equations. Construction of the vascular CFD models is a multi-step process. Critical aspects of the methodology are described in detail including surface reconstruction, construction of the volumetric mesh, imposition of boundary conditions and solution of the finite element model. In vitro and in vivo experimentation is discussed that demonstrates, in a preliminary manner, the validity of the methodology. Flow models are presented for carotid arteries with a wide range of atherosclerotic disease. Considerable evidence has emerged that disturbed blood flow patterns are a major factor in the onset of atherosclerotic disease and may play a role in disease progression. The proposed chapter will review a technique for characterizing blood flow patterns in large arteries from magnetic resonance angiography (MRA) and velocity-encoded phase-contrast magnetic resonance imaging. This technique, known as vascular computational fluid dynamics (CFD), has been applied extensively to the bifurcation region of the carotid artery, a common site of plaque formation. Common hemodynamic features in this region will be presented based on imaging of a series of normal subjects. Hemodynamic features in the vicinity of the carotid bifurcation will also be presented for a series of subjects with advanced atherosclerotic disease.

Efficient pipeline for image-based patient-specific analysis of cerebral aneurysm hemodynamics: technique and sensitivity.

Hemodynamic factors are thought to be implicated in the progression and rupture of intracranial aneurysms. Current efforts aim to study the possible associations of hemodynamic characteristics such as complexity and stability of intra-aneurysmal flow patterns, size and location of the region of flow impingement with the clinical history of aneurysmal rupture. However, there are no reliable methods for measuring blood flow patterns in vivo. In this paper, an efficient methodology for patient-specific modeling and characterization of the hemodynamics in cerebral aneurysms from medical images is described. A sensitivity analysis of the hemodynamic characteristics with respect to variations of several variables over the expected physiologic range of conditions is also presented. This sensitivity analysis shows that although changes in the velocity fields can be observed, the characterization of the intra-aneurysmal flow patterns is not altered when the mean input flow, the flow division, the viscosity model, or mesh resolution are changed. It was also found that the variable that has the greater impact on the computed flow fields is the geometry of the vascular structures. We conclude that with the proposed modeling pipeline clinical studies involving large numbers cerebral aneurysms are feasible.