Encontre seu próximo livro favorito

Torne'se membro hoje e leia gratuitamente por 30 dias.
High Performance Deformable Image Registration Algorithms for Manycore Processors

High Performance Deformable Image Registration Algorithms for Manycore Processors

Ler amostra

High Performance Deformable Image Registration Algorithms for Manycore Processors

Comprimento:
211 página
1 hora
Lançado em:
Jun 28, 2013
ISBN:
9780124078802
Formato:
Livro

Descrição

High Performance Deformable Image Registration Algorithms for Manycore Processors develops highly data-parallel image registration algorithms suitable for use on modern multi-core architectures, including graphics processing units (GPUs). Focusing on deformable registration, we show how to develop data-parallel versions of the registration algorithm suitable for execution on the GPU. Image registration is the process of aligning two or more images into a common coordinate frame and is a fundamental step to be able to compare or fuse data obtained from different sensor measurements. Extracting useful information from 2D/3D data is essential to realizing key technologies underlying our daily lives. Examples include autonomous vehicles and humanoid robots that can recognize and manipulate objects in cluttered environments using stereo vision and laser sensing and medical imaging to localize and diagnose tumors in internal organs using data captured by CT/MRI scans.

  • Demonstrates how to redesign widely used image registration algorithms so as to best expose the underlying parallelism available in these algorithms
  • Shows how to pose and implement the parallel versions of the algorithms within the single instruction, multiple data (SIMD) model supported by GPUs
  • Provides Programming "tricks" that can help readers develop other image processing algorithms, including registration algorithms for the GPU
Lançado em:
Jun 28, 2013
ISBN:
9780124078802
Formato:
Livro

Sobre o autor


Relacionado a High Performance Deformable Image Registration Algorithms for Manycore Processors

Livros relacionados
Artigos relacionados

Amostra do Livro

High Performance Deformable Image Registration Algorithms for Manycore Processors - James Shackleford

1

Introduction

Information in This Chapter:

• Motivation for multicore CPU/GPU implementations

• Applications of deformable registration

• Algorithmic approaches to deformable registration

• Organization of the book

1.1 Introduction

The fundamental step for combining three-dimensional (3D) geometric data is registration, which is the process of aligning two or more images that capture the geometric structure of the same scene, but in their own relative coordinate frames, into a common coordinate frame. The images themselves can be obtained at different times and from different viewpoints, using similar or different imaging modalities. Here, we focus on volumetric registration, where the images are pixel or voxel intensities arranged in a regular grid, and the relative alignment of multiple images must be found. Volumetric registration is often used in biomedical imaging, e.g., to track changes in a patient’s anatomy using images taken at different time points or to align stacks of microscopy data in either space or time.

A registration is called rigid if the motion or change is limited to global rotations and translations, and is called deformable if it includes complex local variations. One of the images is often called the static or reference image and the second image is the moving image, and registration involves spatially transforming the moving image to align with the reference image. When registering medical images, e.g., of a patient’s anatomy taken at different time points, one must account for deformation of the anatomy itself due to the patient’s breathing, anatomical changes, and so on.

Modern imaging techniques such as computed tomography (CT), positron emission tomography (PET), and magnetic resonance imaging (MRI) provide physicians with 3D image volumes of patient anatomy which convey information instrumental in treating a wide range of afflictions. It is often useful to register one image volume to another to understand how patient anatomy has changed over time or to relate image volumes obtained via different imaging techniques. For example, MRI provides a means of distinguishing soft tissues that are otherwise indiscernible in a transmission-based CT scan, and the recent availability of portable CT scanners inside the operating room has led to the development of new methods of localizing cancerous soft tissue by registering intraoperative CT scans to a preoperative MRI as shown in Figure 1.1, thus allowing for precise tumor localization during the resection procedure.

Figure 1.1 Computing organ motion via deformable registration. (A) A preoperative MRI image (in red) superimposed on an intraoperative CT image (in blue) before deformable registration. (B) The preoperative MRI superimposed on the intraoperative CT after deformable registration. (C) The deformation vector field (in blue) derived by the registration process superimposed on the intraoperative CT scan wherein the vector field quantitatively describes the organ motion between the CT and MRI scans.

Efficient and timely processing of 3D data obtained from high-resolution/high-throughput imaging systems requires image analysis algorithms to be significantly accelerated, and registration is no exception. In fact, modern registration algorithms are computationally intensive, and reports of deformable registration algorithms requiring hours to compute for demanding image resolutions and applications are not uncommon (Aylward et al., 2007). Cluster computing is a well-established technique for accelerating image-processing algorithms, since, in many cases, these algorithms can be appropriately parallelized and operations performed independently on different portions of the image. Recent advances in multicore processor design, however, offer new opportunities for truly large-scale and cost-effective parallel computing right at the desk of an individual researcher. For example, CPUs in Intel’s Core i7 family have up to six processing cores operating at 3.5 GHz each, and can achieve a peak processing rate of about 100 GLOPs. Graphics Processing Units (GPUs) are considerably more powerful: a modern GPU such as the NVidia C2050 has 448 cores, each operating at 1.1 GHz, and can achieve a peak processing rate of one TFLOP. However, the processing cores on GPUs are considerably simpler in their design than CPU cores. For algorithms that can be parallelized within its programming model, a single GPU offers the computing power equivalent to a small cluster of CPUs.

This book develops highly data-parallel deformable image registration algorithms suitable for use on modern multicore architectures, including GPUs. Reducing the execution time incurred by modern registration algorithms will allow these techniques to be routinely used in both time-sensitive and data-intensive

Você chegou ao final desta amostra. Inscreva-se para ler mais!
Página 1 de 1

Análises

O que as pessoas pensam sobre High Performance Deformable Image Registration Algorithms for Manycore Processors

0
0 avaliações / 0 Análises
O que você acha?
Classificação: 0 de 5 estrelas

Avaliações de leitores