2D CT图像到3D MRI的转换技术

5星 · 超过95%的资源需积分: 31 146 浏览量更新于2024-09-17 5 收藏 566KB PDF 举报

"该资源是一篇来自IOP Publishing的Physics in Medicine & Biology期刊的文章，发表于2011年，标题为'Registration of 2D x-ray images to 3D MRI by generating pseudo-CT data'，由MJ van der Bom等人撰写。文章探讨了如何将2D的X射线图像注册到3D的MRI图像上，通过生成伪CT数据来实现这一过程。" 在医学图像处理中，2D到3D的转换是一项关键技术，特别是对于CT（计算机断层扫描）和MRI（磁共振成像）这两种不同模态的图像融合。这篇文章研究的核心在于解决2D X射线图像与3D MRI图像之间的配准问题。通常，CT图像提供了良好的解剖结构的密度信息，而MRI图像则擅长显示软组织的细节。然而，由于它们的成像原理不同，直接配准2D X射线图像到3D MRI图像是一项挑战。作者提出的方法是生成伪CT数据，这允许将2D X射线图像的信息转换到与MRI图像相同的三维空间中。这个过程可能包括对2D X射线图像进行一系列的预处理步骤，如图像增强、标准化和分割，然后利用这些信息创建一个模拟CT图像的三维模型。生成的伪CT数据可以作为桥梁，帮助在CT和MRI图像之间建立准确的配准，这对于结合两种图像的优势进行诊断和治疗计划至关重要。配准在医学影像分析中的应用广泛，例如在放射治疗规划、介入手术导航和手术模拟等场景。通过精确的图像配准，医生可以更准确地定位病变、评估疾病进展或治疗效果，从而提高患者护理的质量。此外，这项工作也涉及到图像处理和计算机视觉领域的算法开发，如基于特征的配准方法、变形场估计以及优化算法的应用。这些技术的进步有助于提高配准的精度和速度，减少人为误差，对医学影像学的发展产生了深远影响。这篇研究论文详细探讨了2D到3D图像转换在医学图像配准中的应用，尤其是在将X射线图像与MRI图像相结合的过程中，通过生成伪CT数据解决了不同模态图像之间的融合问题，为临床实践提供了有力的技术支持。

1032 MJvanderBomet al

Introduction

Three-dimensional (3D) anatomical information can be of great value during many image-

guided interventions. However, in most cases only two-dimensional (2D) x-ray projection

images can be acquired during the intervention. For this reason, registration of pre-treatment

3D volume data to 2D x-ray images during clinical procedures is considered beneﬁcial in a

variety of clinical applications. For instance, the registration of portal images to 3D volume

data can provide assistance in treatment planning and patient positioning for radiotherapy

(Bijhold 1993, Gilhuijs et al 1996, Murphy 1997,Siroiset al 1999,Kimet al 2001, Clippe

et al 2003,Jinet al 2006, Jans and Syme 2006, Khamene et al 2006, Munbodh et al 2006,

Chen et al 2008). In the ﬁelds of minimally invasive and computer-aided surgery (Lemieux

et al 1994,Lavall

ee and Szeliski 1995, Feldmar et al 1997, Weese et al 1997,Gu

eziec et al

1998, Kita et al 1998,Liuet al 1998, Maurer et al 1998, Hamadeh et al 1998, Penney et al

1998, 2001, Rohlﬁng and Maurer 2002, Benameur et al 2003,Toma

zevi

c et al 2003, Livyatan

et al 2003, Birkfellner et al 2003,Kimet al 2005, Rohlﬁng et al 2005, Russakoff et al 2005,

Turgeon

et al 2005, McLaughlin et al 2005, Birkfellner et al 2005, Dey and Napel 2006,

Toma

zevi

c et al 2006, Aouadi and Sarry 2008, Fu and Kuduvalli 2008,Markeljet al 2008,van

der Bom et al 2010), navigation of medical devices can be performed using volumetric data

acquired prior to the procedure, offering 3D insight into the patient’s anatomy. A great variety

of 2D–3D registration methods have been developed of which most are applicable only for

the registration of computed tomography (CT) to x-ray images. Because both these imaging

modalities use x-rays, their visual content in terms of tissue contrast is rather similar. These

similarities are well exploited in various registration methods proposed (Bijhold 1993, Gilhuijs

et al 1996, Murphy 1997,Siroiset al 1999,Kimet al 2001, Clippe et al 2003,Jinet al 2006,

Jans and Syme 2006, Khamene et al 2006, Munbodh et al 2006, Chen et al 2008, Lemieux

et al 1994,Lavall

ee and Szeliski 1995, Feldmar et al 1997, Weese et al 1997,Gu

eziec et al

1998, Kita et al 1998,Liuet al 1998, Maurer et al 1998, Hamadeh et al 1998, Penney et al

1998, 2001, Rohlﬁng and Maurer 2002

, Benameur et al 2003, Livyatan et al 2003, Birkfellner

et al 2003,Kimet al 2005, Rohlﬁng et al 2005, Russakoff et al 2005, Turgeon et al 2005,

McLaughlin et al 2005, Birkfellner et al 2005, Dey and Napel 2006, Aouadi and Sarry 2008,

Fu and Kuduvalli 2008, van der Bom et al 2010, Maes et al 1997, Hill et al 2001, Pluim et al

2003).

However, for some applications additional images providing adequate soft tissue contrasts

may be of great value. This soft tissue contrast can be useful during endovascular

interventions like embolization of arteriovenous malformations or orthopedic interventions

like vertebroplasty. Registration of x-ray images to magnetic resonance imaging (MRI) data

may provide such complementary soft tissue contrast which is lacked by x-ray images.

Registering x-ray to MRI data is technically very challenging because fundamental

differences in imaging physics and contrast mechanisms lead to very different tissue contrasts.

Rohlﬁng and Maurer (2002) proposed a similarity measure that uses the distributions of MRI

intensities along rays. Subsequently, these distributions are compared to the pixel intensities

of the x-ray image in a manner that is similar to mutual information. A quantitative evaluation

of this method has so far not been reported. Toma

zevi

c et al (2003) have developed a method

that registers the bony edges of vertebral bodies extracted from 3D volume data to the high

intensity gradients in the x-ray images. Because the method uses gradient information it is

applicable for the registration of x-ray to CT and MR. A drawback of this method is that

the extraction of bone edges in MRI data is not an easy task. Furthermore, Toma

zevi

c et al

propose a method that registers a coarse 3D reconstruction, generated from two or more

oblique x-ray images, to MRI data (Toma

zevi

c et al 2006). Like the gradient-based method

剩余13页未读，继续阅读

Mandy2015

粉丝: 0

2D CT图像到3D MRI的转换技术

OpenCV实现的图像2D转3D Image-2D-to-3D.rar

dicom格式CT图片转换软件

图片2d转3d代码

2dto3d:2D 图像转 3D-matlab开发

MATLAB开发实现2D图像转3D模型技术分享

2D图像转3D动画GIF的汉化转换工具介绍

2D 图像到 STL 网格：将 2D 图像转换为 3D stl CAD 模型-matlab开发

从2D图像进行3D重建

strabo:将HGT高程图转换为2D图像或3D模型

将图像2D转换为3D教程

最新资源