http://niftyrec.scienceontheweb.net/wordpress/
NiftyRec
Welcome to the home page of NiftyRec Open Source Tomography Toolbox, developed at the Centre for Medical Image Computing, University College London.
Versatile:NiftyRec provides reconstruction tools for emission and transmission computed tomography, supporting a wide range of modalities:
§Positron Emission Tomography (PET) –with resolution modelling
§Single Photon Emission Computed Tomography (SPECT) –with depth-dependent resolution modelling
§X-Ray Computed Tomography and Tomosynthesis –cone-beam, fan-beam, helical cone-beam geometries
§Synchrotron X-Ray Tomography –parallel beam geometry
§Neutron Tomography –parallel beam geometry
§Optical Projection Tomography –parallel beam geometry
High performance:NiftyRec is the fastest tomographic reconstruction software out there! Projection, back-projection and core iterative reconstruction routines are highly optimised. Computationally intensive tasks are GPU accelated, achieving unprecedented performance. Compute 3D OSEM SPECT and PET interative reconstructions in 10 seconds!
高性能:
NiftyRec是最快的断层重建软件!投影,反投影和核心迭代重建程序都是高度优化的。计算密集型任务使用GPU加速,达到了前所未有的性能。计算3D OSEM SPECT和PET迭代重建在10秒!
Easy to use:The Matlab and Python interfaces of NiftyRec are intended to enable fast prototyping and development of reconstruction algorithms. The Matlab and Python interfaces include simple demos of standard iterative reconstruction algorithms such as Maximum Likelihood Expectation Maximisation (MLEM), Ordered Subsets Expectation Maximisation (OSEM) and One Step Late Maximum A Posteriori Expectation Maximisation (OSL-MAPEM), applied to PET, SPECT, cone-beam X-Ray CT and parallel-beam X-Ray CT. Other imaging modalities can be easily implemented with a few lines of Matlab and Python.
我的niftyRec笔记01
niftyrec
重建软件最新版本有两个不同版,分别是cuda和no_cuda的。我使用的no_cuda版本,pet,spect(emission
tomography,软件中缩写et_),ct(transmission
Tomography,软件中缩写tt_)都用到了GPU加速,算法主要是MLEM(Maximum Lidelihood
Expectation
Maximization最大似然期望最大),OSEM(有序子集期望最大化),pet,spect的demo算法都可以顺利在matlab里面调用,遇到的问题就是该软件的ct重建demo的三维GPU加速代码不全,缺少一个tt_project_ray_mex函数(注:mex函数是matlab的二进制文件,好像无法打开编辑)。
要使用cuda加速版本的话,计算机没有英伟达的显卡配置也没有问题,因为它有模拟器可以运行GPU计算。
这个软件可以在1:matlab里面使用 addpath 'E:\NiftyRec\niftyrec\matlab'加载进去
,然后直接调用软件自带的各种demo算法就行。
还:2:3D Slicer
plugin插件:即NiftyRec
includes a plugin for 3D Slicer for the reconstruction of SPECT
images.
3:python,Make
sure that the NiftyRec Python extension
module扩展模块is
installedInstall NiftyRec.
注意:Not
all of NiftyRec tools are available in Python at the moment. Take a
look at the PyGTK GUI in the /apps folder in the source code to see
an example of SPECT reconstruction. 又是SPECT!!!
这个软件有三个大的方面:重建,配准以及分割。重建部分分pet,spect和ct。ct(软件中以缩写tt_开头)的重建算法比较少,只有如下图:
而关于pet和spect的很多很细: