TIGRE Windows11+CUDA11.6+Visual Studio 2022 编译安装

TIGRE Windows11+CUDA11.6+Visual Studio 2022 编译安装

TIGRECERN/TIGRE: TIGRE: Tomographic Iterative GPU-based Reconstruction Toolbox (github.com)

介绍

TIGRE: Tomographic Iterative GPU-based Reconstruction Toolbox

TIGRE is an open-source toolbox for fast and accurate 3D tomographic reconstruction for any geometry. Its focus is on iterative algorithms for improved image quality that have all been optimized to run on GPUs (including multi-GPUs) for improved speed. It combines the higher level abstraction of MATLAB or Python with the performance of CUDA at a lower level in order to make it both fast and easy to use.

TIGRE is free to download and distribute: use it, modify it, add to it, share it. Our aim is to provide a wide range of easy-to-use algorithms for the tomographic community “off the shelf”. We would like to build a stronger bridge between algorithm developers and imaging researchers/clinicians by encouraging and supporting contributions from both sides into TIGRE.

TIGRE features

TIGRE is a GPU-based CT reconstruction software repository that contains a wide variety of iterative algorithms.

• MATLAB and Python libraries for high-performance x-ray absorption tomographic reconstruction.
• State-of-the-art implementations of projection and backprojection operations on GPUs (including multi-GPUs), with a simple interface using higher level languages to facilitate the development of new methods.
• Flexible CT geometry: Cone Beam, Parallel Beam, Digital Tomosynthesis, C-arm CT, and any other geometry. Geometric parameters are defined per projection, not per scan.
• A wide range of reconstruction algorithms for CT.
  • Filtered backprojection (FBP,FDK) and variations (different filters, Parker weights, …)
  • Iterative algorithms
    • Gradient-based algorithms (SART, OS-SART, SIRT, ASD-POCS, OS-ASD-POCS, B-ASD-POCS-β, PCSD, AwPCSD, Aw-ASD-POCS) with multiple tuning parameters (Nesterov acceleration, initialization, parameter reduction, …)
    • Krylov subspace algorithms (CGLS, LSQR, hybrid LSQR, LSMR, IRN-TV-CGLS, hybrid-fLSQR-TV, AB/BA-GMRES)
    • Statistical reconstruction (MLEM)
    • Variational methods (FISTA, SART-TV)
• TV denoising for 3D images.
• Basic image loading functionality.
• A variety of plotting functions.
• Image quality metrics.
• Nikon and Varian and Phillips (DICOM) scanner data loaders.

Windows11+CUDA11.6+Visual Studio 2022 安装

1. 将Matlab源码中的 mex_CUDA_win64_MVS2022.xml 文件复制一份重新命名为mex_CUDA_win64.xml
2. 修改 mex_CUDA_win64.xml中，cuda的环境变量，主要是将9.2修改为11.6
3. 在命令行中输入 mex -setup 回车，选择你的MSVC编译器，选择MSVC_2022
4. 在matlab命令行运行 Compile.m 程序。正常情况下就会开始编译了！

可能会出现的错误

• 警告: 不支持选定的编译器 ‘NVIDIA CUDA Compiler’，并且未找到其他受支持的编译器。有关选项，请访问
  https://www.mathworks.com/support/compilers。
• 错误使用 mex nvcc warning : The ‘compute_35’, ‘compute_37’, ‘compute_50’, ‘sm_35’, ‘sm_37’ and ‘sm_50’ architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning). ray_interpolated_projection.cu
  

原因和解决方案

原因matlab未正确检测到CUDA路径，需要你在命令行界面输入设置你的CUDA_PATH。

>> setenv('CUDA_PATH','C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6')
>> getenv('CUDA_PATH')

然后重新运行Compile.m

正确编译界面如下：

检测安装是否正确

1. 运行 InitTIGRE.m 初始话环境

2. 运行Demos文件夹中 d03_generateData.m程序，出现

恭喜成功安装！

20230414，FXXK！！！不行了！！！