0

这是一个关于 vtk 程序的问题。

我正在 3D MRI 和 3D US 图像之间进行严格的配准。美国数据的大小比 MRI 大得多。首先，我在两张图像的同一个器官上取一些地标。然后，我使用“vtkLandmarkTransform”在这些地标点之间进行转换。最后，我使用“vtkImageReslice”将转换应用于 MRI 数据。所以我得到了一个巨大的核磁共振数据。

如何裁剪大量数据以使新的 MRI 数据具有与美国图像相同的大小并显示与美国图像相同的器官位置？

她是我的密码

vtkSmartPointer<vtkPoints> sourcePoints = vtkSmartPointer<vtkPoints>::New();//Input source points
for ( int t = 0; t < 6; t++ )
{
    float sourcePoint[3] = {inital_points_mri[t][0], inital_points_mri[t][1], inital_points_mri[t][2]};
    sourcePoints->InsertNextPoint(sourcePoint);
}

vtkSmartPointer<vtkPoints> targetPoints = vtkSmartPointer<vtkPoints>::New();//Input target points
for ( int t = 0; t < 6; t++ )
{
    float targetPoint[3] = {inital_points_us[t][0], inital_points_us[t][1], inital_points_us[t][2]};
    targetPoints->InsertNextPoint(targetPoint);
}

vtkSmartPointer<vtkLandmarkTransform> landmarkTransform = vtkSmartPointer<vtkLandmarkTransform>::New();//get landmark transform
landmarkTransform->SetSourceLandmarks(sourcePoints);
landmarkTransform->SetTargetLandmarks(targetPoints);
landmarkTransform->SetModeToSimilarity ();
landmarkTransform->Update();

vtkSmartPointer<vtkImageReslice> transform2 = vtkSmartPointer<vtkImageReslice>::New(); // Apply transform
transform2->SetInput(MRIimagedata); // Input MRI image
transform2->AutoCropOutputOn();  // I'm not sure I need it or not?
// transform2->SetOutputExtent(0,499,0,489,0,358); // set the extent as US data, but can't get the same organ position
landmarkTransform->Inverse();
transform2->SetResliceTransform(landmarkTransform);
transform2->Update();
vtkImageData* transformImage = transform2->GetOutput();

感谢您的关注。

4

1 回答 1

0

对不起大家。我在这部分犯了一个愚蠢的错误。

我在原点 (0, 0, 0) 的坐标上获取 MRI 的地标。然后我将变换应用于 MRI 数据，但忘记了 MRI 数据的偏移量不是 (0, 0, 0)。这就是为什么我无法得到正确的结果。

回到我的问题：如何使用 VTK 裁剪图像数据？答案很简单：使用以下代码设置输出原点和扩展参数：

vtkSmartPointer<vtkImageReslice> transform2 = vtkSmartPointer<vtkImageReslice>::New();
transform2->SetInput(MRIimagedata);
// transform2->AutoCropOutputOn();  //It's not necessary 
landmarkTransform1->Inverse();
transform2->SetResliceTransform(landmarkTransform1);
transform2->SetOutputOrigin(0,0,0);  //set the origin of new image data. It's where we start our crop
transform2->SetOutputExtent(0,499,0,489,0,358);  //set the range to crop from big data

如果转换正确，您将得到正确的结果。

这是我的代码

// read MRI image data
vtkSmartPointer<vtkMetaImageReader> reader1 =
    vtkSmartPointer<vtkMetaImageReader>::New();
reader1->SetFileName("MRI.mhd");
reader1->Update();
vtkImageData* MRIimagedata = reader1->GetOutput();
double* origin_MRI = MRIimagedata->GetOrigin();

// generate landmark transforms
vtkSmartPointer<vtkPoints> sourcePoints1 = vtkSmartPointer<vtkPoints>::New();//Input landmarks on MRI data, should add the offset of MRI data
for ( int t = 0; t < 6; t++ )
{
    float sourcePoint1[3] = {inital_points_mri[t][0]+origin_MRI[0], inital_points_mri[t][1]+origin_MRI[1], inital_points_mri[t][2]+origin_MRI[2]};
    sourcePoints1->InsertNextPoint(sourcePoint1);
}

vtkSmartPointer<vtkPoints> targetPoints1 = vtkSmartPointer<vtkPoints>::New();//input landmarks on US data, the offset of US is (0,0,0)
for ( int t = 0; t < 6; t++ )
{
    float targetPoint1[3] = {inital_points_us[t][0], inital_points_us[t][1], inital_points_us[t][2]};
    targetPoints1->InsertNextPoint(targetPoint1);
}

vtkSmartPointer<vtkLandmarkTransform> landmarkTransform1 = vtkSmartPointer<vtkLandmarkTransform>::New();
landmarkTransform1->SetSourceLandmarks(sourcePoints1);
landmarkTransform1->SetTargetLandmarks(targetPoints1);
landmarkTransform1->SetModeToSimilarity();
landmarkTransform1->Update();

vtkSmartPointer<vtkImageReslice> transform2 = vtkSmartPointer<vtkImageReslice>::New();
transform2->SetInput(MRIimagedata);
// transform2->AutoCropOutputOn();
landmarkTransform1->Inverse();
transform2->SetResliceTransform(landmarkTransform1);
transform2->SetOutputOrigin(0,0,0);
transform2->SetOutputExtent(0,499,0,489,0,358);
transform2->Update();
vtkImageData* transformImage = transform2->GetOutput();
于 2013-05-15T15:48:35.120 回答