Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/bunpc/angioreg

register two volumetric angiograms together using an affine transformation then corrected by non-linear scattered data interpolation
https://github.com/bunpc/angioreg

Last synced: 5 months ago
JSON representation

register two volumetric angiograms together using an affine transformation then corrected by non-linear scattered data interpolation

Awesome Lists containing this project

README 

          



# Description:

This GUI is used to register two 2D/3D angiograms by manually selecting matching points between the two angiograms. The Matlab function “procrustes” is used to create the affine transformation given these matching points.This GUI will then perform the transformation of the images.The image transformation is done with our own custom algorithm that is not optimized. We should adopt an optimized algorithm for 3D image transformation. After Linear registartion, images may not align well due to motion atrifacts or other types of non-linear transformations. To correct this, the GUI has two nonlinear regsitration options using the matlab methods "imregdemons" and "scatteredInterpolant". 



# Input: 

Angiogram image files 


Two file names are provided that contain the volumetric image of the two angiograms to be registered. The files can be MAT or TIF files. Please note that MAT files assume the image is ordered as z, x, and y; while the TIF files assume the image is ordered as x, y, and z. See the example files.



# Output: 

The following information can be saved from the GUI.



pts2register.mat contains matching points used for the registration (‘pts2register’ is a user defined variable name, and can be arbitrarily named while saving the points.) 


Pts2register.mat has 2 variables: 


*Angio1pts [Nx3]* - contains 3D coordinates of the points on Angiogram1 


*Angio2pts [Nx3]* - contains 3D coordinates of the points on Angiogram2



Transformation.mat - This contains an array of variables. Not all variables are always created. Each variable contains the following information: 


*DestinationPts [Nx3]* - contains 3D coordinates of the points on Angiogram1 


*SourcePts [Nx3]* - contains 3D coordinates of the points on Angiogram2  


*D2to1 [Z2xX2xY2x3]* - contains the displacement field of Angiogram2  


*Tlist2to1* - is a cell which contains the type of operations performed and the order in which they were performed.  


*DestinationPath* - is the path to the  destination angiogram. 


*SourcePath* - is the path to the source angiogram.



TransformedAngio1.mat contains angiogram1 transformed into the coordinates of angiogram2. The name ‘TransformedAngio1’ is user defined, and can be set while saving the transformed angiogram. 


TransformedAngio1.mat has one variable: 


*angio [Z2xX2xY2]* - contains the transformed angiogram1 which has the same size as angiogram2. Z2, X2 and Y2 are the number of elements in angiogram2 in the Z, X and Y directions respectively.



TransformedAngio2.mat contains angiogram2 transformed into the coordinates of angiogram1. The name ‘TransformedAngio2’ is user defined, and can be set while saving the transformed angiogram. 


TransformedAngio2.mat has 1 variable: 


*angio [Z1xX1xY1]* - contains the transformed angiogram2 which has the same size as angiogram1. Z1, X1 and Y1 are the number of elements in angiogram1 in the Z, X and Y directions respectively.



# Instructions to run angioReg GUI:

* Make sure KuraSuite is in your matlab path using the matlab command setpaths(path, genpath(cd)) from the KuraSuite root directory (or from the angioReg sub-directory).

* Change directory to the folder containing your data files.

* Type angioReg in the MATLAB command window (Do not open by double clicking angioReg.fig. It will open the GUI but it may not function properly).

* To Load the angiograms go to **File > Load Data**

* A window prompt opens. Select the TIF or mat file that contains your first angiogram data.

* Another window prompt will open. Select the second TIF or mat file that contains your OCT or TPM data.

* Rotate 90 Anticlockwise,Rotate 90 clockwise and Transpose push buttons helps to orient the Angiogram2 such that it will be easy to identify identical points.

* If needed, load previously selected points **File > Load points** or else go to the next step.

* Select identical capillary segment bifurcation points on the first image (left axis) and the second image (right axis). To do this, select the Add radio button under image1. Pick a branching point you can easily identify on both images. Then use the Z slider on the right of both images to set your desired Z position for the point. Please set ZMIP to 1 for both volumes to unambiguously identify the z position of the points. Lastly, select the bifurcation point in both volumes with the mouse left click.

* Repeat these steps for at least 10 points. For the best results, scatter the points uniformly throughout the volume. 

* To delete points, select Remove in the button group under image 1 and then click on the number you’d like to remove. A window will appear that says “Would you like to delete point  x ?”. Select “Yes”

* After selecting your points, save the results **File > Save points**. Note that you are advised to regularly save your work as you progress. You don’t need to wait until everything is selected before saving.

* After selecting all desired points, go to the Tools menu to perform the Registration. There, you can choose to **Tools > Linear Registration > Angiogram 1 to 2** if you want to register angiogram1 to angiogram2 or **Tools > Linear Registration > Angiogram 2 to 1** if you want to register angiogram2 to angiogram1.

* Once the registration is complete, you can overlay the transformed image onto the target using the Overlay check box. 

* If Angiogram1 is registered to Angiogram2 then select the Transformed Angiogram1 radio button and select the Overlay radiobutton. Axis1 will display Transformed Angiogram1 with gray color and Angiogram2 with a green color overlay.

* If Angiogram2 is registered to Angiogram1 then select the Transformed Angiogram2 radio button and select the Overlay radiobutton. Axis1 will display Angiogram2 with gray color and Transformed Angiogram1 with a green color overlay.

* This GUI has 2 non-linear registration options if needed. One is using the matlab method imregdemons, the other uses scatteredInterpolant.

* Always try using imregdemons first. To perfrom this non-linear registration using this method go to **Tools > Non Linear Registartion (imregdemons) > Angiogram 1 to 2** if you want to register angiogram 1 to 2 or **Tools > Non Linear Registartion (imregdemons) > Angiogram 2 to 1** if you want to register angiogram 2 to 1.

* For non linear registration using imregdemons, there are a few available options.

  * No histogram equalization

  * Histogram equalization

  * Adaptive histogram equalization

* It is recommended to try all these options to see which works best. 

* A list window prompt will now open. This list will have options to choose for performing the non-linear transformation. These options are

  * Raw data

  * Linear Registered Data (This option will only show if you have already done this registration)

  * Non-Linear registration using scattered Data interpolation (This option will only show if you have already done this registration)

* If the raw data volumes are close to each other then you may not need to do linear registration before doing a non-linear registration.

* If Non-Linear registration using imregdemons doesn't perform well, then try register using scattered data interpolation.

* The non-linear transformation using scattered data interpolation needs many common points to be selected throught out the volume.

* To perform the non-linear registration using this method, go to **Tools > Non Linear Registartion (Scattered Data Interpolation) > Angiogram 1 to 2** if you want to register angiogram 1 to 2, or **Tools > Non Linear Registartion (Scattered Data Interpolation) > Angiogram 2 to 1** if you want to register angiogram 2 to 1.

* If you select one of these two options, a list window prompt will open. This list will have options to choose for performing the non-linear transformation. These options are

  * Raw data

  * Linear Registered Data (This option will only show if you have already did this registration)

  * Non-Linear registration using scattered Data interpolation (This option will only show if you have already did this registration)

* To save the transformed angiograms go to **File > Save Transformed angiogram1** or **File > Save Transformed angiogram2**

* To save the transformation information go to **File > Save Transformation**. This will save the Transformation.mat file in both the source and destination folders.



# Tips

* Always try linear regisration and  non-linear registration (imregdemons) first. If this doesn't give you the results you want then try linnar regisration  and non-linear registration (Scattered data interpolation). 

* If the images are already close enough without linear registration, then Linear registration may not be necessary. Apply the non-linear registraton method directly.

* To register two volumes/images from the same imaging modality, most of the time the non-linear registration (imregdemons) will give you the results you want. To register volumes/images from different imaging modalities, non-linear registration (imregdemons) may not work. In that case, use non-linear registration (Scattered data interpolation).