Hippocampus MRI Parallel Segmentation Using Three Dimensions Lattice Boltzmann Model with Prior Information / 中国医疗器械杂志

Getting volume change of hippocampus by segmenting on brain MRI is an important step in the diagnose of Alzheimer's disease and other brain disease. Three dimensional segmentation can make use of the correlation of image in gray and spatial position, so it has high accuracy. This paper proposes a novel three-dimensional lattice Boltzmann model combined with the surface evolution of deformable model and taking the prior information as an external force term to constrain the evolution of three dimensional surfaces. In order to solve the problem of high computational cost caused by 3D segmentation, the parallelization of the method is programmed on single GPU platform and dual GPU platform. Comparison experiments were set to test the accuracy of segmentation and computational efficiency between the novel LB method and another method by using 20 real AD patient's MRI from ADNI. In ensuring the accuracy of the segmentation, the time can be reduced to 12.76 s on single GPU platform, and 17.32 s on dual GPU platform, contrasting 132.43 s on CPU platform. It fully validates the characteristics of lattice Boltzmann method which can be highly parallelized.

Multi-angle Plane-wave Beamforming Algorithm Based on CUDA / 中国医疗器械杂志

Multi-angle plane-wave beamforming algorithm is the basis of ultra-fast ultrasonic imaging. It can be used to improve the imaging frame rate and resolution of traditional focused ultrasound. However, the existing multi-angle plane-wave technology can not satisfy the real-time imaging requirements due to the huge amount of computation required by CPU. In this paper, We proposed a parallel processing method to reduce the computation time based on compute unified device architecture(CUDA). Simulation analysis and contrast experiment were conducted to verify its performance. Experimental results show that the execution time based on GPU is much less than that based on CPU, thus the computational speed is accelerated significantly to satisfy the demand of ultrafast imaging.

Evaluation of computer methods for biomarker discovery on computational grids

Background: Discovering biomarkers is a fundamental step to understand and deal with genetic diseases. Methods using classic Computer Science algorithms have been adapted in order to support processing large biological data sets, aiming to find useful information to understand causing conditions of diseases such as cancer. Results: This paper describes some promising biomarker discovery methods based on several grid architectures. Each technique has some features that make it more suitable for a particular grid architecture. This matching depends on the parallelizing capabilities of the method and the resource availability in each processing/storage node. Conclusion: The study described in this paper analyzed the performance of biomarker discovery methods in different grid architectures. We find that some methods are more suited for certain grid architectures, resulting in significant performance improvement and producing more accurate results.

Preliminary Study on the Enhancement of Reconstruction Speed for Emission Computed Tomography Using Parallel Processing / 핵의학분자영상

PURPOSE: Conventional image reconstruction uses simplified physical models of projection. However, real physics, for example 3D reconstruction, takes too long time to process all the data in clinic and is unable in a common reconstruction machine because of the large memory for complex physical models. We suggest the realistic distributed memory model of fast-reconstruction using parallel processing on personal computers to enable large-scale technologies. MATERIALS AND METHODS: The preliminary tests for the possibility on virtual manchines and various performance test on commercial super computer, Tachyon were performed. Expectation maximization algorithm with common 2D projection and realistic 3D line of response were tested. Since the process time was getting slower (max 6 times) after a certain iteration, optimization for compiler was performed to maximize the efficiency of parallelization. RESULTS: Parallel processing of a program on multiple computers was available on Linux with MPICH and NFS. We verified that differences between parallel processed image and single processed image at the same iterations were under the significant digits of floating point number, about 6 bit. Double processors showed good efficiency (1.96 times) of parallel computing. Delay phenomenon was solved by vectorization method using SSE. CONCLUSION: Through the study, realistic parallel computing system in clinic was established to be able to reconstruct by plenty of memory using the realistic physical models which was impossible to simplify

Effect of distributed parallel-computing on the time of resampling of digitized human / 第三军医大学学报

Objective To explore the effect of distributed parallel-computing on the time of resampling of digitized human with different amounts of client.Methods Computers were randomly divided into one group of serial-computing and five groups of parallel-computing,and the same data of digitized human was resampled by each group.The resampling time of clients and server in each group,speedup of parallel-computing were compared.Results Compared with the group of serial-computing,the time of resampling of the groups of parallel-computing decreased greatly(P