Automatic Segmentation of the Left Atrium from Computed Tomography Angiography Images.

The left atrial appendage (LAA) causes 91% of thrombi in atrial fibrillation patients, a potential harbinger of stroke. Leveraging computed tomography angiography (CTA) images, radiologists interpret the left atrium (LA) and LAA geometries to stratify stroke risk. Nevertheless, accurate LA segmentation remains a time-consuming task with high inter-observer variability. Binary masks of the LA and their corresponding CTA images were used to train and test a 3D U-Net to automate LA segmentation. One model was trained using the entire unified-image-volume while a second model was trained on regional patch-volumes which were run for inference and then assimilated back into the full volume. The unified-image-volume U-Net achieved median DSCs of 0.92 and 0.88 for the train and test sets, respectively; the patch-volume U-Net achieved median DSCs of 0.90 and 0.89 for the train and test sets, respectively. This indicates that the unified-image-volume and patch-volume U-Net models captured up to 88 and 89% of the LA/LAA boundary's regional complexity, respectively. Additionally, the results indicate that the LA/LAA were fully captured in most of the predicted segmentations. By automating the segmentation process, our deep learning model can expedite LA/LAA shape, informing stratification of stroke risk.

Subject-specific factors affecting particle residence time distribution of left atrial appendage in atrial fibrillation: A computational model-based study.

Background: Atrial fibrillation (AF) is a prevalent arrhythmia, that causes thrombus formation, ordinarily in the left atrial appendage (LAA). The conventional metric of stroke risk stratification, CHA2DS2-VASc score, does not account for LAA morphology or hemodynamics. We showed in our previous study that residence time distribution (RTD) of blood-borne particles in the LAA and its associated calculated variables (i.e., mean residence time, tm , and asymptotic concentration, C ∞) have the potential to improve CHA2DS2-VASc score. The purpose of this research was to investigate the effects of the following potential confounding factors on LAA tm and C ∞: (1) pulmonary vein flow waveform pulsatility, (2) non-Newtonian blood rheology and hematocrit level, and (3) length of the simulation. Methods: Subject-Specific data including left atrial (LA) and LAA cardiac computed tomography, cardiac output (CO), heart rate, and hematocrit level were gathered from 25 AF subjects. We calculated LAA tm and C ∞ based on series of computational fluid dynamics (CFD) analyses. Results: Both LAA tm and C ∞ are significantly affected by the CO, but not by temporal pattern of the inlet flow. Both LAA tm and C ∞ increase with increasing hematocrit level and both calculated indices are higher for non-Newtonian blood rheology for a given hematocrit level. Further, at least 20,000 s of CFD simulation is needed to calculate LAA tm and C ∞ values reliably. Conclusions: Subject-specific LA and LAA geometries, CO, and hematocrit level are essential to quantify the subject-specific proclivity of blood cell tarrying inside LAA in terms of the RTD function.

Numerical Simulation of Concussive-Generated Cortical Spreading Depolarization to Optimize DC-EEG Electrode Spacing for Noninvasive Visual Detection.

BACKGROUND: Cortical spreading depolarization (SD) is a propagating depolarization wave of neurons and glial cells in the cerebral gray matter. SD occurs in all forms of severe acute brain injury, as documented by using invasive detection methods. Based on many experimental studies of mechanical brain deformation and concussion, the occurrence of SDs in human concussion has often been hypothesized. However, this hypothesis cannot be confirmed in humans, as SDs can only be detected with invasive detection methods that would require either a craniotomy or a burr hole to be performed on athletes. Typical electroencephalography electrodes, placed on the scalp, can help detect the possible presence of SD but have not been able to accurately and reliably identify SDs. METHODS: To explore the possibility of a noninvasive method to resolve this hurdle, we developed a finite element numerical model that simulates scalp voltage changes that are induced by a brain surface SD. We then compared our simulation results with retrospectively evaluated data in patients with aneurysmal subarachnoid hemorrhage from Drenckhahn et al. (Brain 135:853, 2012). RESULTS: The ratio of peak scalp to simulated peak cortical voltage, Vscalp/Vcortex, was 0.0735, whereas the ratio from the retrospectively evaluated data was 0.0316 (0.0221, 0.0527) (median [1st quartile, 3rd quartile], n = 161, p < 0.001, one sample Wilcoxon signed-rank test). These differing values provide validation because their differences can be attributed to differences in shape between concussive SDs and aneurysmal subarachnoid hemorrhage SDs, as well as the inherent limitations in human study voltage measurements. This simulated scalp surface potential was used to design a virtual scalp detection array. Error analysis and visual reconstruction showed that 1 cm is the optimal electrode spacing to visually identify the propagating scalp voltage from a cortical SD. Electrode spacings of 2 cm and above produce distorted images and high errors in the reconstructed image. CONCLUSIONS: Our analysis suggests that concussive (and other) SDs can be detected from the scalp, which could confirm SD occurrence in human concussion, provide concussion diagnosis on the basis of an underlying physiological mechanism, and lead to noninvasive SD detection in the setting of severe acute brain injury.

Length of Stay in the Neonatal ICU is Predictable using Heart Rate: An Opportunity for Optimizing Managed Care.

We explore the use of classification and regression models for predicting the length of stay (LoS) of neonatal patients in the intensive care unit (ICU), using heart rate (HR) time-series data of 7,758 patients from the MIMIC-IH database. We find that aggregated features of HR on the first full-day of in-patient stay after admission (i.e. the first day with a full 24-hour record for each patient) can be leveraged to classify LoS in excess of 10 days with 89% sensitivity and 59% specificity. As such, LoS as a continuous variable was also found to be statistically significantly correlated to aggregate HR data corresponding to the first full-day after admission.

Modeling Cases and Deaths per Million using Daily-Aggregated Facebook COVID-19 Symptom Survey Data.

We develop a novel analytic approach to modeling future COVID-19 risk using COVID-19 Symptom Survey data aggregated daily by US state, joined with daily time-series data on confirmed cases and deaths. Specifically, we model N-day forward-looking estimates for per-US-state-per-day change in deaths per million (DPM) and cases per million (CPM) using a multivariate regression model to below baseline error (65% and 38% mean absolute percentage error for DPM/CPM, respectively). Additionally, we model future changes in the curvature of CPM/DPM as "increasing" or "decreasing" using a random forest classifier to above 72% accuracy. In sum, we develop and characterize models to establish a relationship between behaviors and beliefs of individuals captured via the Facebook COVID-19 Symptom Surveys and the trajectory of COVID-19 outbreaks evidenced in terms of CPM and DPM. Such information can be helpful in assessing collective risks of infection and death during a pandemic as well as in determining the effectiveness of appropriate risk mitigation strategies based on behaviors evidenced through survey responses.

Intelligent patient monitoring for proactive alerting of key personnel in intensive care: A single-center study.

A code blue event is an emergency code to indicate when a patient goes into cardiac arrest and needs resuscitation. In this paper, we model the binary response of a intensive care unit (ICU) patients experiencing a code-blue event, starting with vital time-series data of patients in 12 ICU beds. Our study introduces day-of and day-ahead risk scoring models trained against ground truth information on per-patient-per-day code-blue events, starting with multi-variate vital-time-series-sequences of varying durations with a plurality of engineered features capturing temporal variations of these signals. Actionable events, including code-blue events, aggregated by patient by day were predicted on the day-of or day-ahead with an overall accuracy of over 80% in our best models. Such models have potential to improve healthcare delivery by providing just-in-time alerting, enabling proactive and preventative clinical interventions, through continuous patient monitoring.

Convolutional Neural Network based Segmentation of Abdominal Aortic Aneurysms.

Abdominal aortic aneurysms (AAAs) are balloonlike dilations in the descending aorta associated with high mortality rates. Between 2009 and 2019, reported ruptured AAAs resulted in ~28,000 deaths while reported unruptured AAAs led to ~15,000 deaths. Automating identification of the presence, 3D geometric structure, and precise location of AAAs can inform clinical risk of AAA rupture and timely interventions. We investigate the feasibility of automatic segmentation of AAAs, inclusive of the aorta, aneurysm sac, intra-luminal thrombus, and surrounding calcifications, using 30 patient-specific computed tomography angiograms (CTAs). Binary masks of the AAA and their corresponding CTA images were used to train and test a 3D U-Net - a convolutional neural network (CNN) - model to automate AAA detection. We also studied model-specific convergence and overall segmentation accuracy via a loss-function developed based on the Dice Similarity Coefficient (DSC) for overlap between the predicted and actual segmentation masks. Further, we determined optimum probability thresholds (OPTs) for voxel-level probability outputs of a given model to optimize the DSC in our training set, and utilized 3D volume rendering with the visualization tool kit (VTK) to validate the same and inform the parameter optimization exercise. We examined model-specific consistency with regard to improving accuracy by training the CNN with incrementally increasing training samples and examining trends in DSC and corresponding OPTs that determine AAA segmentations. Our final trained models consistently produced automatic segmentations that were visually accurate with train and test set losses in inference converging as our training sample size increased. Transfer learning led to improvements in DSC loss in inference, with the median OPT of both the training segmentations and testing segmentations approaching 0.5, as more training samples were utilized.

A method for the identification of COVID-19 biomarkers in human breath using Proton Transfer Reaction Time-of-Flight Mass Spectrometry.

BACKGROUND: COVID-19 has caused a worldwide pandemic, making the early detection of the virus crucial. We present an approach for the determination of COVID-19 infection based on breath analysis. METHODS: A high sensitivity mass spectrometer was combined with artificial intelligence and used to develop a method for the identification of COVID-19 in human breath within seconds. A set of 1137 positive and negative subjects from different age groups, collected in two periods from two hospitals in the USA, from 26 August, 2020 until 15 September, 2020 and from 11 September, 2020 until 11 November, 2020, was used for the method development. The subjects exhaled in a Tedlar bag, and the exhaled breath samples were subsequently analyzed using a Proton Transfer Reaction Time-of-Flight Mass Spectrometer (PTR-ToF-MS). The produced mass spectra were introduced to a series of machine learning models. 70% of the data was used for these sub-models' training and 30% was used for testing. FINDINGS: A set of 340 samples, 95 positives and 245 negatives, was used for the testing. The combined models successfully predicted 77 out of the 95 samples as positives and 199 out of the 245 samples as negatives. The overall accuracy of the model was 81.2%. Since over 50% of the total positive samples belonged to the age group of over 55 years old, the performance of the model in this category was also separately evaluated on 339 subjects (170 negative and 169 positive). The model correctly identified 166 out of the 170 negatives and 164 out of the 169 positives. The model accuracy in this case was 97.3%. INTERPRETATION: The results showed that this method for the identification of COVID-19 infection is a promising tool, which can give fast and accurate results.

The establishment of a 3D anatomical coordinate system for defining vaginal axis and spatial position.

BACKGROUND AND OBJECTIVE: Pelvic organ prolapse (POP), the herniation of the pelvic organs toward the vaginal opening, is a common pelvic floor disorder (PFD) whose etiology is poorly understood. Traditional methods for evaluating POP are often constrained to external vaginal examination, limited to 2D, or have poor reproducibility. We propose a reliable 3D anatomic coordinate system for standardized 3D assessment of pelvic anatomy using magnetic resonance imaging (MRI). METHODS: The novel 3D anatomic reference system is based on six bony landmarks of the pelvis manually identified in MRI: the ischial spines and the superior and inferior pubic points of the left and right pubic symphysis. The origin of this system is defined as the midpoint of the ischial spines. The reproducibility and applicability of the pelvic coordinate system were evaluated by (1) implementing it in a new method to quantify vaginal position and axis (angulation) in 3D space from MRI segmentations of the vagina and (2) computing the intraclass correlation (ICC) on coordinate system and vaginal measures. The MRI analysis was performed by four non-medically trained observers on five pelvic MRI datasets on approximately five separate occasions. RESULTS: Overall, all bony landmarks had excellent intra-observer reliability and inter-observer reliability (ICC>0.90); intra-observer reliability was moderate-to-good among the vaginal position parameters (0.5

Subject-Specific Calculation of Left Atrial Appendage Blood-Borne Particle Residence Time Distribution in Atrial Fibrillation.

Atrial fibrillation (AF) is the most common arrhythmia that leads to thrombus formation, mostly in the left atrial appendage (LAA). The current standard of stratifying stroke risk, based on the CHA2DS2-VASc score, does not consider LAA morphology, and the clinically accepted LAA morphology-based classification is highly subjective. The aim of this study was to determine whether LAA blood-borne particle residence time distribution and the proposed quantitative index of LAA 3D geometry can add independent information to the CHA2DS2-VASc score. Data were collected from 16 AF subjects. Subject-specific measurements included left atrial (LA) and LAA 3D geometry obtained by cardiac computed tomography, cardiac output, and heart rate. We quantified 3D LAA appearance in terms of a novel LAA appearance complexity index (LAA-ACI). We employed computational fluid dynamics analysis and a systems-based approach to quantify residence time distribution and associated calculated variable (LAA mean residence time, t m) in each subject. The LAA-ACI captured the subject-specific LAA 3D geometry in terms of a single number. LAA t m varied significantly within a given LAA morphology as defined by the current subjective method and it was not simply a reflection of LAA geometry/appearance. In addition, LAA-ACI and LAA t m varied significantly for a given CHA2DS2-VASc score, indicating that these two indices of stasis are not simply a reflection of the subjects' clinical status. We conclude that LAA-ACI and LAA t m add independent information to the CHA2DS2-VASc score about stasis risk and thereby can potentially enhance its ability to stratify stroke risk in AF patients.

Association of low-voltage areas with the regional wall deformation and the left atrial shape in patients with atrial fibrillation: A proof of concept study.

BACKGROUND: Left atrium (LA) remodeling is associated with atrial fibrillation (AF) and reduced success after AF ablation, but its relation with low-voltage areas (LVA) is not known. This study aimed to evaluate the relation between regional LA changes and LVAs in AF patients. METHODS: Pre-interventional CT data of patients (n = 24) with LA-LVA (<0.5 mV) in voltage mapping after AF ablation were analyzed (Surgery Explorer, QuantMD LLC). To quantify asymmetry (ASI = LA-A/LAV) a cutting plane parallel to the rear wall and along the pulmonary veins divided the LA-volume (LAV) into anterior (LA-A) and posterior parts. To quantify sphericity (LAS = 1-R/S), a patient-specific best-fit LA sphere was created. The average radius (R) and the mean deviation (S) from this sphere were calculated. The average local deviation (D) was measured for the roof, posterior, septum, inferior septum, inferior-posterior and lateral walls. RESULTS: The roof, posterior and septal regions had negative local deviations. There was a correlation between roof and septum (r = 0.42, p = 0.04), lateral and inferior-posterior (r = 0.48, p = 0.02) as well as posterior and inferior-septal deviations (r = -0.41, p = 0.046). ASI correlated with septum deformation (r = -0.43, p = 0.04). LAS correlated with dilatation (LAV, r = 0.49, p = 0.02), roof (r = 0.52, p = 0.009) and posterior deformation (r = -0.56, p = 0.005). Extended LVA correlated with local deformation of all LA walls, except the roof and the septum. LVA association with LAV, ASI and LAS did not reach statistical significance. CONCLUSION: Extended LVA correlates with local wall deformations better than other remodeling surrogates. Therefore, their calculation could help predict LVA presence and deserve further evaluation in clinical studies.

Morphological Analysis of the Right Ventricular Endocardial Wall in Pulmonary Hypertension.

Pulmonary hypertension (PH) is a chronic progressive disease diagnosed when the pressure in the main pulmonary artery, assessed by right heart catheterization (RHC), is greater than 25 mmHg. Changes in the pulmonary vasculature due to the high pressure yield an increase in the right ventricle (RV) afterload. This starts a remodeling process during which the ventricle exhibits changes in shape and eventually fails. RV models were obtained from the segmentation of cardiac magnetic resonance images at baseline and 1-year follow-up for a pilot study that involved 12 PH and 7 control subjects. The models were used to create surface meshes of the geometry and to compute the principal, mean, and Gaussian curvatures. Ten global curvature indices were calculated for each of the RV endocardial wall reconstructions at the end-diastolic volume (EDV) and end-systolic volume (ESV) phases of the cardiac cycle. Statistical analysis of the data was performed to discern if there are significant differences in the curvature indices between controls and the PH group, as well as between the baseline and follow-up phases for the PH subjects. Six curvature indices, namely, the Gaussian curvature at ESV, the mean curvature at EDV and ESV, the L2-norm of the mean curvature at ESV, and the L2-norm of the major principal curvature at EDV and ESV, were found to be significantly different between controls and PH subjects (p < 0.05). We infer that these geometry measures could be used as indicators of RV endocardial wall morphology changes. Two global parameters, the Gaussian and mean curvatures at ESV, showed significant changes at the one-year follow-up for the PH subjects (p < 0.05). The aforementioned geometry measures to assess changes in RV shape could be used as part of a noninvasive computational tool to aid clinicians in PH diagnostic and progression assessment, and to evaluate the effectiveness of treatment.

Design, Development, and Temporal Evaluation of a Magnetic Resonance Imaging-Compatible In Vitro Circulation Model Using a Compliant Abdominal Aortic Aneurysm Phantom.

Biomechanical characterization of abdominal aortic aneurysms (AAAs) has become commonplace in rupture risk assessment studies. However, its translation to the clinic has been greatly limited due to the complexity associated with its tools and their implementation. The unattainability of patient-specific tissue properties leads to the use of generalized population-averaged material models in finite element analyses, which adds a degree of uncertainty to the wall mechanics quantification. In addition, computational fluid dynamics modeling of AAA typically lacks the patient-specific inflow and outflow boundary conditions that should be obtained by nonstandard of care clinical imaging. An alternative approach for analyzing AAA flow and sac volume changes is to conduct in vitro experiments in a controlled laboratory environment. In this study, we designed, built, and characterized quantitatively a benchtop flow loop using a deformable AAA silicone phantom representative of a patient-specific geometry. The impedance modules, which are essential components of the flow loop, were fine-tuned to ensure typical intraluminal pressure conditions within the AAA sac. The phantom was imaged with a magnetic resonance imaging (MRI) scanner to acquire time-resolved images of the moving wall and the velocity field inside the sac. Temporal AAA sac volume changes lead to a corresponding variation in compliance throughout the cardiac cycle. The primary outcome of this work was the design optimization of the impedance elements, the quantitative characterization of the resistive and capacitive attributes of a compliant AAA phantom, and the exemplary use of MRI for flow visualization and quantification of the deformed AAA geometry.

A Comparative Classification Analysis of Abdominal Aortic Aneurysms by Machine Learning Algorithms.

The objective of this work was to perform image-based classification of abdominal aortic aneurysms (AAA) based on their demographic, geometric, and biomechanical attributes. We retrospectively reviewed existing demographics and abdominal computed tomography angiography images of 100 asymptomatic and 50 symptomatic AAA patients who received an elective or emergent repair, respectively, within 1-6 months of their last follow up. An in-house script developed within the MATLAB computational platform was used to segment the clinical images, calculate 53 descriptors of AAA geometry, and generate volume meshes suitable for finite element analysis (FEA). Using a third party FEA solver, four biomechanical markers were calculated from the wall stress distributions. Eight machine learning algorithms (MLA) were used to develop classification models based on the discriminatory potential of the demographic, geometric, and biomechanical variables. The overall classification performance of the algorithms was assessed by the accuracy, area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and precision of their predictions. The generalized additive model (GAM) was found to have the highest accuracy (87%), AUC (89%), and sensitivity (78%), and the third highest specificity (92%), in classifying the individual AAA as either asymptomatic or symptomatic. The k-nearest neighbor classifier yielded the highest specificity (96%). GAM used seven markers (six geometric and one biomechanical) to develop the classifier. The maximum transverse dimension, the average wall thickness at the maximum diameter, and the spatially averaged wall stress were found to be the most influential markers in the classification analysis. A second classification analysis revealed that using maximum diameter alone results in a lower accuracy (79%) than using GAM with seven geometric and biomechanical markers. We infer from these results that biomechanical and geometric measures by themselves are not sufficient to discriminate adequately between population samples of asymptomatic and symptomatic AAA, whereas MLA offer a statistical approach to stratification of rupture risk by combining demographic, geometric, and biomechanical attributes of patient-specific AAA.

Real-World Variability in the Prediction of Intracranial Aneurysm Wall Shear Stress: The 2015 International Aneurysm CFD Challenge.

PURPOSE: Image-based computational fluid dynamics (CFD) is widely used to predict intracranial aneurysm wall shear stress (WSS), particularly with the goal of improving rupture risk assessment. Nevertheless, concern has been expressed over the variability of predicted WSS and inconsistent associations with rupture. Previous challenges, and studies from individual groups, have focused on individual aspects of the image-based CFD pipeline. The aim of this Challenge was to quantify the total variability of the whole pipeline. METHODS: 3D rotational angiography image volumes of five middle cerebral artery aneurysms were provided to participants, who were free to choose their segmentation methods, boundary conditions, and CFD solver and settings. Participants were asked to fill out a questionnaire about their solution strategies and experience with aneurysm CFD, and provide surface distributions of WSS magnitude, from which we objectively derived a variety of hemodynamic parameters. RESULTS: A total of 28 datasets were submitted, from 26 teams with varying levels of self-assessed experience. Wide variability of segmentations, CFD model extents, and inflow rates resulted in interquartile ranges of sac average WSS up to 56%, which reduced to < 30% after normalizing by parent artery WSS. Sac-maximum WSS and low shear area were more variable, while rank-ordering of cases by low or high shear showed only modest consensus among teams. Experience was not a significant predictor of variability. CONCLUSIONS: Wide variability exists in the prediction of intracranial aneurysm WSS. While segmentation and CFD solver techniques may be difficult to standardize across groups, our findings suggest that some of the variability in image-based CFD could be reduced by establishing guidelines for model extents, inflow rates, and blood properties, and by encouraging the reporting of normalized hemodynamic parameters.

Improvements in left ventricular twist mechanics following myectomy for hypertrophic cardiomyopathy with mid-ventricular obstruction.

We aimed to evaluate left ventricle twist mechanics in mid-ventricular obstructive and apical type of hypertrophic cardiomyopathy and changes induced by myectomy. We studied 3 consecutive patients by cardiac magnetic resonance preoperatively and 6 weeks after myectomy. We calculated the apical and basal rotations at the base and apex respectively. All 3 patients underwent myectomy by the standard described technique. The basal rotations remained the same, while there was an improvement in the maximal apical rotation from 0.385 ± 0.3975° to 0.9086 ± 1.1751°. In hypertrophic cardiomyopathy with mid-ventricular obstruction and apical hypertrophy, there is decreased apical rotation, which improves after myectomy.

Improvements in Regional Left Ventricular Function Following Late Percutaneous Coronary Intervention for Anterior Myocardial Infarction.

BACKGROUND: Late revascularization following a myocardial infarction has questionable clinical benefit. METHODS: We studied 13 patients with anterior wall myocardial infarction who underwent percutaneous coronary intervention within 2 weeks of the primary event, by quantitative analysis of 2-dimensional echocardiographic images. Endocardial segmentations of the left ventricular (LV) endocardium from the 4-chamber views were studied over time to establish cumulative wall displacements (CWDs) throughout the cardiac cycle. RESULTS: Left ventricular end-systolic volume decreased to 42 ± 8 mL/body surface area (P = .034) and LV ejection fraction improved to 52% ± 7% (P = .04). Analysis of LV endocardial CWD demonstrated significant improvements in mid-systolic to late-systolic phases in the apical LV segments, from 3.5 ± 0.32 to 5.89 ± 0.43 mm (P = .019). Improvements in CWD were also observed in the late-diastolic phase of the cardiac cycle, from 1.50 ± 0.42 to 1.76 ± 0.52 mm (P = .04). CONCLUSIONS: In our pilot patient cohort, following late establishment of infarct-related artery patency following an anterior wall myocardial infarction, regional improvements were noted in the LV apical segments during systole and late diastole.

Alk1 controls arterial endothelial cell migration in lumenized vessels.

Heterozygous loss of the arterial-specific TGFß type I receptor, activin receptor-like kinase 1 (ALK1; ACVRL1), causes hereditary hemorrhagic telangiectasia (HHT). HHT is characterized by development of fragile, direct connections between arteries and veins, or arteriovenous malformations (AVMs). However, how decreased ALK1 signaling leads to AVMs is unknown. To understand the cellular mis-steps that cause AVMs, we assessed endothelial cell behavior in alk1-deficient zebrafish embryos, which develop cranial AVMs. Our data demonstrate that alk1 loss has no effect on arterial endothelial cell proliferation but alters arterial endothelial cell migration within lumenized vessels. In wild-type embryos, alk1-positive cranial arterial endothelial cells generally migrate towards the heart, against the direction of blood flow, with some cells incorporating into endocardium. In alk1-deficient embryos, migration against flow is dampened and migration in the direction of flow is enhanced. Altered migration results in decreased endothelial cell number in arterial segments proximal to the heart and increased endothelial cell number in arterial segments distal to the heart. We speculate that the consequent increase in distal arterial caliber and hemodynamic load precipitates the flow-dependent development of downstream AVMs.

The Computational Fluid Dynamics Rupture Challenge 2013--Phase II: Variability of Hemodynamic Simulations in Two Intracranial Aneurysms.

With the increased availability of computational resources, the past decade has seen a rise in the use of computational fluid dynamics (CFD) for medical applications. There has been an increase in the application of CFD to attempt to predict the rupture of intracranial aneurysms, however, while many hemodynamic parameters can be obtained from these computations, to date, no consistent methodology for the prediction of the rupture has been identified. One particular challenge to CFD is that many factors contribute to its accuracy; the mesh resolution and spatial/temporal discretization can alone contribute to a variation in accuracy. This failure to identify the importance of these factors and identify a methodology for the prediction of ruptures has limited the acceptance of CFD among physicians for rupture prediction. The International CFD Rupture Challenge 2013 seeks to comment on the sensitivity of these various CFD assumptions to predict the rupture by undertaking a comparison of the rupture and blood-flow predictions from a wide range of independent participants utilizing a range of CFD approaches. Twenty-six groups from 15 countries took part in the challenge. Participants were provided with surface models of two intracranial aneurysms and asked to carry out the corresponding hemodynamics simulations, free to choose their own mesh, solver, and temporal discretization. They were requested to submit velocity and pressure predictions along the centerline and on specified planes. The first phase of the challenge, described in a separate paper, was aimed at predicting which of the two aneurysms had previously ruptured and where the rupture site was located. The second phase, described in this paper, aims to assess the variability of the solutions and the sensitivity to the modeling assumptions. Participants were free to choose boundary conditions in the first phase, whereas they were prescribed in the second phase but all other CFD modeling parameters were not prescribed. In order to compare the computational results of one representative group with experimental results, steady-flow measurements using particle image velocimetry (PIV) were carried out in a silicone model of one of the provided aneurysms. Approximately 80% of the participating groups generated similar results. Both velocity and pressure computations were in good agreement with each other for cycle-averaged and peak-systolic predictions. Most apparent "outliers" (results that stand out of the collective) were observed to have underestimated velocity levels compared to the majority of solutions, but nevertheless identified comparable flow structures. In only two cases, the results deviate by over 35% from the mean solution of all the participants. Results of steady CFD simulations of the representative group and PIV experiments were in good agreement. The study demonstrated that while a range of numerical schemes, mesh resolution, and solvers was used, similar flow predictions were observed in the majority of cases. To further validate the computational results, it is suggested that time-dependent measurements should be conducted in the future. However, it is recognized that this study does not include the biological aspects of the aneurysm, which needs to be considered to be able to more precisely identify the specific rupture risk of an intracranial aneurysm.

cMRI-BED: A novel informatics framework for cardiac MRI biomarker extraction and discovery applied to pediatric cardiomyopathy classification.

BACKGROUND: Pediatric cardiomyopathies are a rare, yet heterogeneous group of pathologies of the myocardium that are routinely examined clinically using Cardiovascular Magnetic Resonance Imaging (cMRI). This gold standard powerful non-invasive tool yields high resolution temporal images that characterize myocardial tissue. The complexities associated with the annotation of images and extraction of markers, necessitate the development of efficient workflows to acquire, manage and transform this data into actionable knowledge for patient care to reduce mortality and morbidity. METHODS: We develop and test a novel informatics framework called cMRI-BED for biomarker extraction and discovery from such complex pediatric cMRI data that includes the use of a suite of tools for image processing, marker extraction and predictive modeling. We applied our workflow to obtain and analyze a dataset of 83 de-identified cases and controls containing cMRI-derived biomarkers for classifying positive versus negative findings of cardiomyopathy in children. Bayesian rule learning (BRL) methods were applied to derive understandable models in the form of propositional rules with posterior probabilities pertaining to their validity. Popular machine learning methods in the WEKA data mining toolkit were applied using default parameters to assess cross-validation performance of this dataset using accuracy and percentage area under ROC curve (AUC) measures. RESULTS: The best 10-fold cross validation predictive performance obtained on this cMRI-derived biomarker dataset was 80.72% accuracy and 79.6% AUC by a BRL decision tree model, which is promising from this type of rare data. Moreover, we were able to verify that mycocardial delayed enhancement (MDE) status, which is known to be an important qualitative factor in the classification of cardiomyopathies, is picked up by our rule models as an important variable for prediction. CONCLUSIONS: Preliminary results show the feasibility of our framework for processing such data while also yielding actionable predictive classification rules that can augment knowledge conveyed in cardiac radiology outcome reports. Interactions between MDE status and other cMRI parameters that are depicted in our rules warrant further investigation and validation. Predictive rules learned from cMRI data to classify positive and negative findings of cardiomyopathy can enhance scientific understanding of the underlying interactions among imaging-derived parameters.