Krzysztof Jurczuk

Computational modeling of MR flow imaging by the lattice Boltzmann method and Bloch equation

In this work, a computational model of magnetic resonance (MR) flow imaging is proposed. The first model component provides fluid dynamics maps by applying the lattice Boltzmann method. The second one uses the flow maps and couples MR imaging (MRI) modeling with a new magnetization transport algorithm based on the Eulerian coordinate approach. MRI modeling is based on the discrete time solution of the Bloch equation by analytical local magnetization transformations (exponential scaling and rotations). Model is validated by comparison of experimental and simulated MR images in two three-dimensional geometries (straight and U-bend tubes) with steady flow under comparable conditions. Two-dimensional geometries, presented in literature, were also tested. In both cases, a good agreement is observed. Quantitative analysis shows in detail the model accuracy. Computational time is noticeably lower to prior works. These results demonstrate that the discrete time solution of Bloch equation coupled with the new magnetization transport algorithm naturally incorporates flow influence in MRI modeling. As a result, in the proposed model, no additional mechanism (unlike in prior works) is needed to consider flow artifacts, which implies its easy extensibility. In combination with its low computational complexity and efficient implementation, the model could have a potential application in study of flow disturbances (in MRI) in various conditions and in different geometries.

Evolutionary induction of a decision tree for large-scale data: a GPU-based approach

Evolutionary induction of decision trees is an emerging alternative to greedy top-down approaches. Its growing popularity results from good prediction performance and less complex output trees. However, one of the major drawbacks associated with the application of evolutionary algorithms is the tree induction time, especially for large-scale data. In the paper, we design and implement a graphics processing unit (GPU)-based parallelization of evolutionary induction of decision trees. We apply a Compute Unified Device Architecture programming model, which supports general-purpose computation on a GPU (GPGPU). The selection and genetic operators are performed sequentially on a CPU, while the evaluation process for the individuals in the population is parallelized. The data-parallel approach is applied, and thus, the parts of a dataset are spread over GPU cores. Each core processes the assigned chunk of the data. Finally, the results from all GPU cores are merged and the sought tree metrics are sent to the CPU. Computational performance of the proposed approach is validated experimentally on artificial and real-life datasets. A comparison with the traditional CPU version shows that evolutionary induction of decision trees supported by GPGPU can be accelerated significantly (even up to 800 times) and allows for processing of much larger datasets.

In Silico Modeling of Magnetic Resonance Flow Imaging in Complex Vascular Networks

The paper presents a computational model of magnetic resonance (MR) flow imaging. The model consists of three components. The first component is used to generate complex vascular structures, while the second one provides blood flow characteristics in the generated vascular structures by the lattice Boltzmann method. The third component makes use of the generated vascular structures and flow characteristics to simulate MR flow imaging. To meet computational demands, parallel algorithms are applied in all the components. The proposed approach is verified in three stages. In the first stage, experimental validation is performed by an in vitro phantom. Then, the simulation possibilities of the model are shown. Flow and MR flow imaging in complex vascular structures are presented and evaluated. Finally, the computational performance is tested. Results show that the model is able to reproduce flow behavior in large vascular networks in a relatively short time. Moreover, simulated MR flow images are in accordance with the theoretical considerations and experimental images. The proposed approach is the first such an integrative solution in literature. Moreover, compared to previous works on flow and MR flow imaging, this approach distinguishes itself by its computational efficiency. Such a connection of anatomy, physiology and image formation in a single computer tool could provide an in silico solution to improving our understanding of the processes involved, either considered together or separately.

A Parallel Approach for Evolutionary Induced Decision Trees. MPI+OpenMP Implementation

One of the important and still not fully addressed issues in evolving decision trees is the induction time, especially for large datasets. In this paper, the authors propose a parallel implementation for Global Decision Tree system that combines shared memory (OpenMP) and message passing (MPI) paradigms to improve the speed of evolutionary induction of decision tree. The proposed solution is based on the classical master-slave model. The population is evenly distributed to available nodes and cores, and the time consuming operations like fitness evaluation and genetic operators are executed in parallel on slaves. Only the selection is performed on the master node. Efficiency and scalability of the proposed implementation is validated experimentally on artificial datasets. It shows noticeable speedup and possibility to efficiently process large datasets.

Vascular System Modeling in Parallel Environment - Distributed and Shared Memory Approaches

This paper presents two approaches in parallel modeling of vascular system development in internal organs. In the first approach, new parts of tissue are distributed among processors and each processor is responsible for perfusing its assigned parts of tissue to all vascular trees. Communication between processors is accomplished by passing messages, and therefore, this algorithm is perfectly suited for distributed memory architectures. The second approach is designed for shared memory machines. It parallelizes the perfusion process during which individual processing units perform calculations concerning different vascular trees. The experimental results, performed on a computing cluster and multicore machines, show that both algorithms provide a significant speedup.

Hybrid Parallelization of Evolutionary Model Tree Induction

This paper illustrates a parallel implementation of evolutionary induction of model trees. An objective is to demonstrate that such evolutionary evolved trees, which are emerging alternatives to the greedy top-down solutions, can be successfully applied to large scale data. The proposed approach combines message passing (MPI) and shared memory (OpenMP) paradigms. This hybrid approach is based on a classical master-slave model in which the individuals from the population are evenly distributed to available nodes and cores. The most time consuming operations like recalculation of the regression models in the leaves as well as the fitness evaluation and genetic operators are executed in parallel on slaves. Experimental validation on artificial and real-life datasets confirms the efficiency of the proposed implementation.

GPU-based computational modeling of magnetic resonance imaging of vascular structures

Magnetic resonance imaging (MRI) is one of the most important diagnostic tools in modern medicine. Since it is a high-cost and highly-complex imaging modality, computational models are frequently built to enhance its understanding as well as to support further development. However, such models often have to be simplified to complete simulations in a reasonable time. Thus, the simulations with high spatial/temporal resolutions, with any motion consideration (like blood flow) and/or with 3D objects usually call for using parallel computing environments. In this paper, we propose to use graphics processing units (GPUs) for fast simulations of MRI of vascular structures. We apply a CUDA environment which supports general purpose computation on GPU (GPGPU). The data decomposition strategy is applied and thus the parts of each virtual object are spread over the GPU cores. The GPU cores are responsible for calculating the influence of blood flow behavior and MRI events after successive time steps. In the proposed approach, different data layouts, memory access patterns, and other memory improvements are applied to efficiently exploit GPU resources. Computational performance is thoroughly validated for various vascular structures and different NVIDIA GPUs. Results show that MRI simulations can be accelerated significantly thanks to GPGPU. The proposed GPU-based approach may be easily adopted in the modeling of other flow related phenomena like perfusion, diffusion or transport of contrast agents.

Evolutionary Induction of Classification Trees on Spark

Evolutionary-based approaches have recently been increasingly proposed for data mining tasks, but their real applicability depends on efficiency and scalability for large-scale data. It is clear that parallel and distributed processing support is indispensable herein. Apache Spark is one of the most promising cluster-computing engines for Big Data. In this paper, we investigate the application of Spark to speed up an evolutionary induction of classification trees in the Global Decision Tree (GDT) system. The system simultaneously searches for the tree structure and tests in non-terminal nodes due to specialized genetic operators. As the original GDT system is implemented in C++, the Java-based module is developed for Spark-based acceleration of the most computationally demanding fitness evaluation. The training dataset is transformed to Resilient Distributed Dataset, which enables in-memory processing of dataset’s parts on workers. Preliminary experimental validation on large-scale artificial and real-life datasets shows that the proposed solution is efficient and scales well.

GPU Accelerated Simulations of Magnetic Resonance Imaging of Vascular Structures

Computer simulations of magnetic resonance imaging (MRI) are important tools in improving this imaging modality and developing new imaging techniques. However, MRI models often have to be simplified to enable simulations to be carried out in a reasonable time. The computational complexity associated with the tracking of magnetic fields’ perturbations with high spatial and temporal resolutions is very high and, thus, it calls for using parallel computing environments. In this paper, we present a GPU-based parallel approach to simulate MRI of vascular structures. The magnetization calculation in different spatial coordinates is spread over GPU cores. We apply CUDA framework and take advantage of GPU memory hierarchy to efficiently exploit GPU computational power. Experimental results with different GPUs and various images show that the proposed algorithm substantially speedups the simulation. The proposed GPU-based approach may be easily adopted in modeling of other flow related phenomena like perfusion or diffusion.

Hierarchical parallel approach in vascular network modeling: hybrid MPI+OpenMP implementation

This paper presents a two-level parallel algorithm of vascular network development. At the outer level, tasks (newly appeared parts of tissue) are spread over processing nodes. Each node attempts to connect/disconnect its assigned parts of tissue in all vascular trees. Communication between nodes is accomplished by a message passing paradigm. At the inner level, subtasks concerning different vascular trees (e.g. arterial and venous) within each task are parallelized by a shared address space paradigm. The solution was implemented on a computing cluster of multi-core nodes with mixed MPI+OpenMP support. The experimental results show that the algorithm provides a significant improvement in computational performance compared with a pure MPI implementation.