multimodal alignment e.g. anatomic+parametric images

[xav69ier]

Dears - I will be generating parametric images (MRI perfusion) to be aligned to classical anatomic MRI images. I wonder how can I assess (metrics?) the co-registration and eventually decide upon fine tuning of the parameters in the command line, if necessary e.g. for greedy.exe - kind rgds xavier

[sarthakpati]

Hi @xav69ier,

You can access all registration parameters through the command line: https://cbica.github.io/CaPTk/preprocessing_reg.html

There are the following options for registration:

[  -reg, --registration]              The kind of registration to perform
                                     Defaults to Affine
                                     Can use Mask File with '-m' and multiple moving images with '-i'
                                     For Affine, the second number defines the degrees of freedom, eg: '-ref Affine-12'

[  -rFI, --regFixedImg]               The Fixed Image for the registration
                                     Needed for registration

[  -rME, --regMetrics]                The kind of metrics to use: SSD (Sum of Squared Differences) or MI (Mutual Information) or
                                     NMI (Normalized Mutual Information) or NCC-AxBxC (Normalized Cross correlation with integer radius for 3D image)
                                     Defaults to NMI

[  -rNI, --regNoIters]                The number of iterations per level of multi-res
                                     Defaults to 100,50,5

[  -rIS, --regInterSave]              Whether the intermediate files are to be saved or not
                                     Defaults to 0

[  -rSg, --regSegMoving]              Whether the Moving Image(s) is a segmentation file
                                     If 1, the 'Nearest Label' Interpolation is applied
                                     Defaults to 0

[  -rJC, --regJacobian]               Whether Jacobian will be generated or not

[  -rIA, --regInterAffn]              The path to the affine transformation to apply to moving image
                                     If this is present, the Affine registration step will be skipped
                                     Also used for rigid transformation

[  -rID, --regInterDefm]              The path to the deformable transformation to apply to moving image
                                     If this is present, the Deformable registration step will be skipped

[xav69ier]

Thank you @sarthakpati I understand these parameters apply to preprocessing.exe. But my question rather refers to metrics that I may use to evaluate the result of the co-resgistration further than just common visual check.

[sarthakpati]

Oh, apologies, I didn't understand the question!

We don't have any metrics the quantify registration, at the moment. If you have ideas, we are happy to explore them further along with any future work we are planning. You are also welcome to put forth a PR if you have something implemented and we give you the appropriate acknowledgement. 😄

[xav69ier]

we are thankful for the tool, if we come with idea addressing my concern above we will definitively share with you all, best x