Coregistration

Coregistration is the process of aligning two or more medical images into a common coordinate system so that corresponding anatomical structures coincide. It is frequently used to bring images from different imaging modalities, such as magnetic resonance imaging (MRI), computed tomography (CT), and positron emission tomography (PET), into spatial correspondence with one another, or to align images acquired at different time points.

Coregistration relies on estimating a spatial transformation that maps coordinates from a moving image to a reference image. Transformations can be rigid (translation and rotation), affine (including scaling and shearing), or non-linear to accommodate local deformations. The similarity between images is quantified with metrics such as mutual information for multi-modal data, or cross-correlation and sum of squared differences for same-modality data. The transformation is applied through interpolation to resample the moving image into the reference space.

Typical workflow includes preprocessing to correct biases and align brain masks, selecting an appropriate reference image, optimizing the transformation under a chosen metric, and assessing alignment quality. Quality assessment may involve visual inspection and quantitative measures like overlap of labeled regions or similarity scores.

Coregistration is essential in neuroimaging, functional mapping, radiology, and longitudinal studies. It enables accurate voxel-wise comparisons, fusion of anatomical and functional information, and consistent ROI analysis across sessions. Challenges include differences in image resolution, geometric distortion (notably in echo-planar imaging), partial coverage, and interpolation artifacts, which can bias downstream analyses if not carefully addressed.

Common software packages include FSL, ANTs, SPM, and 3D Slicer.