Identification of Vortex Cores in Cerebral Aneurysms on 4D Flow MRI

Influence of vortical structures on fibrin clot formation in cerebral aneurysms: A two-dimensional computational study

Thrombosis is an important contributor to cerebral aneurysm growth and progression. A number of sophisticated multiscale and multiphase in silico models have been developed with a view towards interventional planning. Many of these models are able to account for clotting outcomes, but do not provide detailed insight into the role of flow during clot development. In this study, we present idealised, two-dimensional in silico cerebral fibrin clot model based on computational fluid dynamics (CFD), biochemical modelling and variable porosity, permeability, and diffusivity. The model captures fibrin clot growth in cerebral aneurysms over a period at least 1000 s in five different geometries. The fibrin clot growth results were compared to an experiment presented in literature. The biochemistry was found to be more sensitive to mesh size compared to the haemodynamics, while larger timesteps overpredicted clot size in pulsatile flow. When variable diffusivity was used, the predicted clot size was 25.4% lesser than that with constant diffusivity. The predicted clot size in pulsatile flow was 14.6% greater than in plug flow. Different vortex modes were observed in plug and pulsatile flow; the latter presented smaller intermediate modes where the main vortex was smaller and less likely to disrupt the growing fibrin clot. Furthermore, smaller vortex modes were seen to support fibrin clot propagation across geometries. The model clearly demonstrates how the growing fibrin clot alters vortical structures within the aneurysm sac and how this changing flow, in turn, shapes the growing fibrin clot.

SRflow: Deep learning based super-resolution of 4D-flow MRI data

Exploiting 4D-flow magnetic resonance imaging (MRI) data to quantify hemodynamics requires an adequate spatio-temporal vector field resolution at a low noise level. To address this challenge, we provide a learned solution to super-resolve in vivo 4D-flow MRI data at a post-processing level. We propose a deep convolutional neural network (CNN) that learns the inter-scale relationship of the velocity vector map and leverages an efficient residual learning scheme to make it computationally feasible. A novel, direction-sensitive, and robust loss function is crucial to learning vector-field data. We present a detailed comparative study between the proposed super-resolution and the conventional cubic B-spline based vector-field super-resolution. Our method improves the peak-velocity to noise ratio of the flow field by 10 and 30% for in vivo cardiovascular and cerebrovascular data, respectively, for 4 × super-resolution over the state-of-the-art cubic B-spline. Significantly, our method offers 10x faster inference over the cubic B-spline. The proposed approach for super-resolution of 4D-flow data would potentially improve the subsequent calculation of hemodynamic quantities.

Intracranial Flow Velocity Quantification Using Non-Contrast Four-Dimensional Flow MRI: A Prospective Comparative Study with Transcranial Doppler Ultrasound

Four-dimensional (4D) flow magnetic resonance imaging (MRI) allows three-dimensional velocity encoding to measure blood flow in a single scan, regardless of the intracranial artery direction. We compared blood flow velocity quantification by non-contrast 4D flow MRI and by transcranial Doppler ultrasound (TCD), the most widely used modality for measuring velocity. Twenty-two patients underwent both TCD and non-contrast 4D flow MRI. The mean time interval between TCD and non-contrast 4D flow MRI was 0.7 days. Subsegmental velocities were measured bilaterally in the middle cerebral and basilar arteries using TCD and non-contrast 4D flow MRI. Intracranial velocity measurements using TCD and non-contrast 4D flow MRI demonstrated a strong correlation in the bilateral M1, especially at the proximal segment (right r = 0.74, left r = 0.78; all p < 0.001). Mean velocities acquired with 4D flow MRI were approximately 8 to 10% lower than those acquired with TCD according to the location of M1. Intracranial arterial flow measurements estimated using non-contrast 4D flow MRI and TCD showed strong correlation. 4D flow MRI enables simultaneous assessment of vascular morphology and quantitative hemodynamic measurement, providing three-dimensional blood flow visualization. 4D flow MRI is a clinically useful sequence with a promising role in cerebrovascular disease.

Dynamic modes of inflow jet in brain aneurysms

The transition of the inflow jet to turbulence is crucial in understanding the pathology of brain aneurysms. Previous works Le et al. (2010, 2013) have shown evidence for a highly dynamic inflow jet in the ostium of brain aneurysms. While it is highly desired to investigate this inflow jet dynamics in clinical practice, the constraints on spatial and temporal resolutions of in vivo data do not allow a detailed analysis of this transition. In this work, Dynamic Mode Decomposition (DMD) is used to identify the most energetic modes of the inflow jet in patient-specific models of internal carotid aneurysms via the utilization of high-resolution simulation data. It is hypothesized that dynamic modes are not solely controlled by the blood flow waveform at the parent artery. They are also dependent on jet-wall interaction phenomena. DMD analysis shows that the spatial extent of low- frequency modes corresponds well to the most energetic areas of the inflow jet. The high-frequency modes are short-lived and correspond to the flow separation at the proximal neck and the jet's impingement onto the aneurysmal wall. Low-frequency modes can be reconstructed at relatively low spatial and temporal resolutions comparable to ones of in vivo data. The current results suggest that DMD can be practically useful in analyzing blood flow patterns of brain aneurysms with in vivo data.