Difference between revisions of "Projects:RegistrationLibrary:RegLib C03"

From NAMIC Wiki
Jump to: navigation, search
Line 35: Line 35:
 
*[[Image:Button_red_fixed_white.jpg|20px]]reference/fixed : T2w FSE,  
 
*[[Image:Button_red_fixed_white.jpg|20px]]reference/fixed : T2w FSE,  
 
*[[Image:Button_green_moving_white.jpg|20px]] moving: Baseline image of acquired DTI volume, corresponds to T2w MRI , 0.9375 x 0.9375 x 1.4 mm voxel size, oblique
 
*[[Image:Button_green_moving_white.jpg|20px]] moving: Baseline image of acquired DTI volume, corresponds to T2w MRI , 0.9375 x 0.9375 x 1.4 mm voxel size, oblique
*[[Image:Button_green_moving_purple.jpg|20px]] moving: Tensor data of DTI volume, oblique, same orientation as Baseline image. The result Xform will be applied to this volume.
+
*[[Image:Button_blue_tag_white.jpg|20px]] tag: Tensor data of DTI volume, oblique, same orientation as Baseline image. The result Xform will be applied to this volume.
  
 
=== Registration Results===
 
=== Registration Results===
Line 62: Line 62:
  
 
=== Discussion: Registration Challenges ===
 
=== Discussion: Registration Challenges ===
*accuracy is the critical criterion here. We need the registration error (residual misalignment) to be smaller than the change we want to measure/detect. Agreement on what constitutes good alignment can therefore vary greatly.
+
*The DTI contains acquisition-related distortions (commonly EPI acquisitions) that can make automated registration difficult.
*the two images have strong differences in coil inhomogeneity. This affects less the registration quality but hampers evaluation. Most of the difference does not become apparent until after registration in direct juxtaposition. Bias field correction beforehand is recommended.  
+
*the two images often have strong differences in voxel sizes and voxel anisotropy. If the orientation of the highest resolution is not the same in both images, finding a good match can be difficult.
*we have slightly different voxel sizes
+
*there may be widespread and extensive pathology (e.g stroke, tumor) that might affect the registration if its contrast is different in the baseline and structural reference scan
*if the pathology change is substantial it might affect the registration.
 
  
 
=== Discussion: Key Strategies ===
 
=== Discussion: Key Strategies ===
*the two images have identical contrast, hence we consider "sharper" cost functions, such as NormCorr or MeanSqrd
+
*the two images have identical contrast, hence we could consider "sharper" cost functions, such as NormCorr or MeanSqrd. But because of the strong distortions and lower resolution of the moving image, Mutual Information is recommended as the most robust metric.
*general practice is to register the follow-up to the baseline. However here the follow-up has slightly higher resolution. From an image quality/data perspective it would be better to use the highest resolution image as your fixed/reference. But here follow the most common convention, i.e. fixed image is the baseline.
+
*often anatomical labels are available from the reference scan. It would be less work to align the anatomical reference with the DTI, since that would circumvent having to resample the complex tensor data into a new orientation. However the strong distortions are better addressed by registering the other direction, i.e. move the DTI into the anatomical reference space.
 
*because we seek to assess/quantify regional size change, we must use a rigid (6DOF) scheme, i.e. we must exclude scaling.
 
*because we seek to assess/quantify regional size change, we must use a rigid (6DOF) scheme, i.e. we must exclude scaling.
*if the pathology change is soo large that it might affect the registration, we should mask it out. The simplest way to do this is to build a box ROI from the ROItool and feed it as input to the registration. Remember that masking does not mean that masked areas aren't matched, they just do not contribute to the cost function driving the registration, but move along passively.  Next more involved level would be to outline the tumor. If a segmentation is available, you can use that.
+
*masking is likely necessary to obtain good results.  
 +
*in this example the initial alignment of the two scans is very poor. The strongly oblique orientation of the DTI makes an initial manual alignment step necessary.
 
*these two images are not too far apart initially, so we reduce the default of expected translational misalignment
 
*these two images are not too far apart initially, so we reduce the default of expected translational misalignment
*because accuracy is more important than speed here, we increase the sampling rate from the default 2% to 15%.
+
*because speed is not that critical, we increase the sampling rate from the default 2% to 15%.
*we also expect minimal differences in scale & distortion: so we can either set the expected values to 0 or run a rigid registration
+
*we also expect larger differences in scale & distortion than with regular structural scane: so we significantly  (2x-3x) increase the expected values for scale and skew from the defaults.  
*we test the result in areas with good anatomical detail and contrast, far away from the pathology. With rigid body motion a local measure of registration accuracy is representative and can give us a valid limit of detectable change.
+
*a good affine alignment is important before proceeding to non-rigid alignment to further correct for distortions.

Revision as of 20:45, 25 January 2010

Home < Projects:RegistrationLibrary:RegLib C03

Back to ARRA main page
Back to Registration main page
Back to Registration Use-case Inventory

Slicer Registration Library Exampe #3: Diffusion Weighted Image Volume: align with structural reference MRI

this is the fixed reference image. All images are aligned into this space lleft this is the moving image. The transform is calculated by matching this to the reference image this is a passive image to which the calculated transform is applied. It is a label-map in the same space as the moving FLAIR image LEGEND

lleft this indicates the reference image that is fixed and does not move. All other images are aligned into this space and resolution
lleft this indicates the moving image that determines the registration transform.
lleft this indicates images that passively move into the reference space, i.e. they have the transform applied but do not contribute to the calculation of the transform.

lleft T2 lleft DTI Baseline lleft DTI volume
1mm isotropic
256 x 256 x 146
RAS
1.2mm isotropic
256 x 256 x 116
RAS
1.2mm isotropic
256 x 256 x 116
RAS

Objective / Background

This is a typical example of DTI processing. Goal is to align the DTI image with a structural scan that provides accuracte anatomical reference. The DTI contains acquisition-related distortion and insufficient contrast to discern anatomical detail.

Keywords

MRI, brain, head, intra-subject, DTI, DWI

Input Data

  • Button red fixed white.jpgreference/fixed : T2w FSE,
  • Button green moving white.jpg moving: Baseline image of acquired DTI volume, corresponds to T2w MRI , 0.9375 x 0.9375 x 1.4 mm voxel size, oblique
  • Button blue tag white.jpg tag: Tensor data of DTI volume, oblique, same orientation as Baseline image. The result Xform will be applied to this volume.

Registration Results

after affine alignment

Download


Discussion: Registration Challenges

  • The DTI contains acquisition-related distortions (commonly EPI acquisitions) that can make automated registration difficult.
  • the two images often have strong differences in voxel sizes and voxel anisotropy. If the orientation of the highest resolution is not the same in both images, finding a good match can be difficult.
  • there may be widespread and extensive pathology (e.g stroke, tumor) that might affect the registration if its contrast is different in the baseline and structural reference scan

Discussion: Key Strategies

  • the two images have identical contrast, hence we could consider "sharper" cost functions, such as NormCorr or MeanSqrd. But because of the strong distortions and lower resolution of the moving image, Mutual Information is recommended as the most robust metric.
  • often anatomical labels are available from the reference scan. It would be less work to align the anatomical reference with the DTI, since that would circumvent having to resample the complex tensor data into a new orientation. However the strong distortions are better addressed by registering the other direction, i.e. move the DTI into the anatomical reference space.
  • because we seek to assess/quantify regional size change, we must use a rigid (6DOF) scheme, i.e. we must exclude scaling.
  • masking is likely necessary to obtain good results.
  • in this example the initial alignment of the two scans is very poor. The strongly oblique orientation of the DTI makes an initial manual alignment step necessary.
  • these two images are not too far apart initially, so we reduce the default of expected translational misalignment
  • because speed is not that critical, we increase the sampling rate from the default 2% to 15%.
  • we also expect larger differences in scale & distortion than with regular structural scane: so we significantly (2x-3x) increase the expected values for scale and skew from the defaults.
  • a good affine alignment is important before proceeding to non-rigid alignment to further correct for distortions.