Deep Learning for Time Averaged Wall Shear Stress Prediction in Left Main Coronary Bifurcations

Gharleghi, Ramtin; Samarasinghe, Gihan; Sowmya, Arcot; Beier, Susann

doi:10.1109/isbi45749.2020.9098715

Cited by 16 publications

(13 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous studies have shown neural network's capability to estimate hemodynamic variables from geometric features ( 15 – 18 ). While these methods were able to estimate hemodynamic variables with sufficient accuracy, the estimated values produced were time-averaged or specific to a static boundary condition, as no other quantities (i.e., pressure, velocity) were provided as input values.…”

Section: Discussionmentioning

confidence: 99%

“…Liang et al ( 18 ) also leveraged a geometric approach, with the predicted results showing similar aortic stress distributions, though it was not quantified and was only tested against finite element models. Another study combined vessel diameter and curvature information, showing good time-averaged WSS predictions in coronary arteries, with pattern similarities compared through visual inspection ( 15 ). Pattern matching is typically performed by checking areas of overlap using bins, categorizing WSS magnitudes from low to high ( 14 ).…”

Section: Discussionmentioning

confidence: 99%

“…Recent developments in medical imaging and deep learning have enabled the use of physics-based simulations as surrogates to train a deep learning model ( 15 – 18 ). These approaches offer high accuracy compared to the CFD ground truth by learning spatial representation of geometric features.…”

Section: Introductionmentioning

confidence: 99%

“…Similarly, Acebes et al ( 16 ) presented a CNN-based network to estimate endothelial cell activation potential (ECAP) using an unwrapped model of the left atria. Gharleghi et al ( 15 ) also presented a deep learning method to estimate TAWSS in left main coronary bifurcations by using geometric information as the input. Conversely, conventional curve fitting methods have used velocity information at constant spatial locations (equidistant inward normal).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

WSSNet: Aortic Wall Shear Stress Estimation Using Deep Learning on 4D Flow MRI

Ferdian

Dubowitz

Mauger

et al. 2022

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

Wall shear stress (WSS) is an important contributor to vessel wall remodeling and atherosclerosis. However, image-based WSS estimation from 4D Flow MRI underestimates true WSS values, and the accuracy is dependent on spatial resolution, which is limited in 4D Flow MRI. To address this, we present a deep learning algorithm (WSSNet) to estimate WSS trained on aortic computational fluid dynamics (CFD) simulations. The 3D CFD velocity and coordinate point clouds were resampled into a 2D template of 48 × 93 points at two inward distances (randomly varied from 0.3 to 2.0 mm) from the vessel surface (“velocity sheets”). The algorithm was trained on 37 patient-specific geometries and velocity sheets. Results from 6 validation and test cases showed high accuracy against CFD WSS (mean absolute error 0.55 ± 0.60 Pa, relative error 4.34 ± 4.14%, 0.92 ± 0.05 Pearson correlation) and noisy synthetic 4D Flow MRI at 2.4 mm resolution (mean absolute error 0.99 ± 0.91 Pa, relative error 7.13 ± 6.27%, and 0.79 ± 0.10 Pearson correlation). Furthermore, the method was applied on in vivo 4D Flow MRI cases, effectively estimating WSS from standard clinical images. Compared with the existing parabolic fitting method, WSSNet estimates showed 2–3 × higher values, closer to CFD, and a Pearson correlation of 0.68 ± 0.12. This approach, considering both geometric and velocity information from the image, is capable of estimating spatiotemporal WSS with varying image resolution, and is more accurate than existing methods while still preserving the correct WSS pattern distribution.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

WSSNet: Aortic Wall Shear Stress Estimation Using Deep Learning on 4D Flow MRI

Ferdian

Dubowitz

Mauger

et al. 2022

Front. Cardiovasc. Med.

View full text Add to dashboard Cite

show abstract

“…The drastic acceleration of the computational time by ML is deemed more beneficial if applied to more time-consuming 3D simulations, whose output can be more informative as described earlier. However, the number of attempts made on predicting 3D-simulation-based FFR using ML is limited, although prediction of 3D WSS can be found 43 . The time-consuming nature to generate sufficient training dataset with 3D simulation is a potential reason behind that, although this could be a worthy investment considering the benefit of obtaining FFR (and relevant) data equivalent to 3D simulation output in a very short time.…”

Section: Challenges and New Application Areasmentioning

confidence: 99%

CT-based fractional flow reserve: development and expanded application

Torii

Yacoub

2021

gcsp

View full text Add to dashboard Cite

Computations of fractional flow reserve, based on CT coronary angiography and computational fluid dynamics (CT-based FFR) to assess the severity of coronary artery stenosis, was introduced around a decade ago and is now one of the most successful applications of computational fluid dynamic modelling in clinical practice. Although the mathematical modelling framework behind this approach and the clinical operational model vary, its clinical efficacy has been demonstrated well in general. In this review, technical elements behind CT-based FFR computation are summarised with some key assumptions and challenges. Examples of these challenges include the complexity of the model (such as blood viscosity and vessel wall compliance modelling), whose impact has been debated in the research. Efforts made to address the practical challenge of processing time are also reviewed. Then, further application areas – myocardial bridge, renal stenosis and lower limb stenosis – are discussed along with specific challenges expected in these areas.

show abstract

Mesh Convolutional Neural Networks for Wall Shear Stress Estimation in 3D Artery Models

Suk

Haan

Lippe

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Computational fluid dynamics (CFD) is a valuable tool for personalised, non-invasive evaluation of hemodynamics in arteries, but its complexity and time-consuming nature prohibit large-scale use in practice. Recently, the use of deep learning for rapid estimation of CFD parameters like wall shear stress (WSS) on surface meshes has been investigated. However, existing approaches typically depend on a hand-crafted re-parametrisation of the surface mesh to match convolutional neural network architectures. In this work, we propose to instead use mesh convolutional neural networks that directly operate on the same finiteelement surface mesh as used in CFD. We train and evaluate our method on two datasets of synthetic coronary artery models with and without bifurcation, using a ground truth obtained from CFD simulation. We show that our flexible deep learning model can accurately predict 3D WSS vectors on this surface mesh. Our method processes new meshes in less than 5 [s], consistently achieves a normalised mean absolute error of ≤1.6 [%], and peaks at 90.5 [%] median approximation accuracy over the held-out test set, comparing favourably to previously published work. This demonstrates the feasibility of CFD surrogate modelling using mesh convolutional neural networks for hemodynamic parameter estimation in artery models.

show abstract

Deep Learning for Time Averaged Wall Shear Stress Prediction in Left Main Coronary Bifurcations

Cited by 16 publications

References 13 publications

WSSNet: Aortic Wall Shear Stress Estimation Using Deep Learning on 4D Flow MRI

WSSNet: Aortic Wall Shear Stress Estimation Using Deep Learning on 4D Flow MRI

CT-based fractional flow reserve: development and expanded application

Mesh Convolutional Neural Networks for Wall Shear Stress Estimation in 3D Artery Models

Contact Info

Product

Resources

About