A Fluid-Structure Interaction Analysis of Blood Clot Motion in a Branch of Pulmonary Arteries

Advancing Pulmonary Embolism Detection with Integrated Deep Learning Architectures

The main aim of this study is to introduce a new hybrid deep learning model for biomedical image classification. We propose a novel convolutional neural network (CNN), named HybridNeXt, for detecting pulmonary embolism (PE) from computed tomography (CT) images. To evaluate the HybridNeXt model, we created a new dataset consisting of two classes: (1) PE and (2) control. The HybridNeXt architecture combines different advanced CNN blocks, including MobileNet, ResNet, ConvNeXt, and Swin Transformer. We specifically designed this model to combine the strengths of these well-known CNNs. The architecture also includes stem, downsampling, and output stages. By adjusting the parameters, we developed a lightweight version of HybridNeXt, suitable for clinical use. To further improve the classification performance and demonstrate transfer learning capability, we proposed a deep feature engineering (DFE) method using a multilevel discrete wavelet transform (MDWT). This DFE model has three main phases: (i) feature extraction from raw images and wavelet bands, (ii) feature selection using iterative neighborhood component analysis (INCA), and (iii) classification using a k-nearest neighbors (kNN) classifier. We first trained HybridNeXt on the training images, creating a pretrained HybridNeXt model. Then, using this pretrained model, we extracted features and applied the proposed DFE method for classification. The HybridNeXt model achieved a test accuracy of 90.14%, while our DFE model improved accuracy to 96.35%. Overall, the results confirm that our HybridNeXt architecture is highly accurate and effective for biomedical image classification. The presented HybridNeXt and HybridNeXt-based DFE methods can potentially be applied to other image classification tasks.

"What makes blood clots break off?" A Back-of-the-Envelope Computation Toward Explaining Clot Embolization

PURPOSE

One in four deaths worldwide is due to thromboembolic disease; that is, one in four people die from blood clots first forming and then breaking off or embolizing. Once broken off, clots travel downstream, where they occlude vital blood vessels such as those of the brain, heart, or lungs, leading to strokes, heart attacks, or pulmonary embolisms, respectively. Despite clots' obvious importance, much remains to be understood about clotting and clot embolization. In our work, we take a first step toward untangling the mystery behind clot embolization and try to answer the simple question: "What makes blood clots break off?"

METHODS

To this end, we conducted experimentally-informed, back-of-the-envelope computations combining fracture mechanics and phase-field modeling. We also focused on deep venous clots as our model problem.

RESULTS

Here, we show that of the three general forces that act on venous blood clots-shear stress, blood pressure, and wall stretch-induced interfacial forces-the latter may be a critical embolization force in occlusive and non-occlusive clots, while blood pressure appears to play a determinant role only for occlusive clots. Contrary to intuition and prior reports, shear stress, even when severely elevated, appears unlikely to cause embolization.

CONCLUSION

This first approach to understanding the source of blood clot bulk fracture may be a critical starting point for understanding blood clot embolization. We hope to inspire future work that will build on ours and overcome the limitations of these back-of-the-envelope computations.

In silico analysis of embolism in cerebral arteries using fluid-structure interaction method

Ischemic stroke, particularly embolic stroke, stands as a significant global contributor to mortality and long-term disabilities. This paper presents a comprehensive simulation of emboli motion through the middle cerebral artery (MCA), a prevalent site for embolic stroke. Our patient-specific computational model integrates major branches of the middle cerebral artery reconstructed from magnetic resonance angiography images, pulsatile flow dynamics, and emboli of varying geometries, sizes, and material properties. The fluid-structure interactions method is employed to simulate deformable emboli motion through the middle cerebral artery, allowing observation of hemodynamic changes in artery branches upon embolus entry. We investigated the impact of embolus presence on shear stress magnitude on artery walls, analyzed the effects of embolus material properties and geometries on embolus trajectory and motion dynamics within the middle cerebral artery. Additionally, we evaluated the non-Newtonian behavior of blood, comparing it with Newtonian blood behavior. Our findings highlight that embolus geometry significantly influences trajectory, motion patterns, and hemodynamics within middle cerebral artery branches. Emboli with visco-hyperelastic material properties experienced higher stresses upon collision with artery walls compared to those with hyperelastic properties. Furthermore, considering blood as a non-Newtonian fluid had notable effects on emboli stresses and trajectories within the artery, particularly during collisions. Notably, the maximum von Mises stress experienced in our study was 21.83 kPa, suggesting a very low probability of emboli breaking during movement, impact, and after coming to a stop. However, in certain situations, the magnitude of shear stress on them exceeded 1 kPa, increasing the likelihood of cracking and disintegration. These results serve as an initial step in anticipating critical clinical conditions arising from arterial embolism in the middle cerebral artery. They provide insights into the biomechanical parameters influencing embolism, contributing to improved clinical decision-making for stroke management.