This tutorial demonstrates how to pass data between 3D volumes, based upon their geometric relationship. This method can be used to map volumes of different orientations, sizes, and resolutions. It is also a means of up- or down-sampling a volume.
Introduction
There are various reasons why one might want to transfer data between two volumes, which differ in geometry or resolution. One example might be to map a discrete parcellation (set of labelled regions) to a functional image, for the purpose of averaging the functional signal by region. Another might be simply to achieve a lower or higher resolution image. This tutorial walks you through this data transfer process.
Loading the data
First off, download the tutorial data here, which consists of three volumes:
- suit_cerebellum.nii The "spatially unbiased atlas template of the cerebellum and brainstem" (SUIT) atlas, a set of labelled regions defining the cerebellum and brainstem.
- fdg_pet.nii An FDG-PET image, representing the rate of brain glucose metabolism
- t1.nii A corresponding T1-weighted anatomical image
These volumes are all registered in the same (MNI stereotaxic) space, but have different orientations, sizes, and resolutions. Our task is to map them all to one Volume3D object, visualize the results, and save the image. One immediate application of this is the ability to mask the FDG-PET image by specific cerebellar regions, and use this mask to compute an average cerebellar signal, which can then be used to normalize the signal ().