A 3D Computerized Brain Atlas

In a collaboration with the departments of Clinical Neurophysiology and Neurology at the Karolinska Institute/Hospital, Stockholm,  a Computerized Brain Atlas (CBA) has been developed. CBA, is used to improve quantification and evaluation of neuroimaging data. It uses a detailed three-dimensional brain atlas that can be adapted to fit the individual brain of a patient Anatomical information from the atlas can then be introduced into the images. Furthermore, by applying the inverse atlas transformation, images from a patient can be transformed to conform to the anatomy of the atlas brain. This intersubject registration allows for comparisons of data from different individuals on a pixel-by-pixel basis. CBA can also be used for co-registration of data from different examinations of the same patient.

Surface renderring of the atlas anatomy

Matching of the 3D Atlas to a patient's brain

Matching of the atlas can be made interactively, fully automatic or in a semi-automatic way. For matching, a polynomial transformation with 27 parameters is used. It can account for individual variations in position, size and shape (see the atypical PET-brain below).

  • MR images before (upper row) and after (lower row) interactive matching of the atlas

  • PET data before (upper row) and after (lower row) fully automatic matching of the atlas

    Interrogation of the atlas data base

    The atlas contains 3D descriptions of about 400 structures. Structures can be selected by their names and will then be drawn into displayed images, or the user can point and click in the images.

    The image shows all Brodmann areas outlined in a PET-slice (left), and the identification of a single gyrus in an MR-slice using the point-and-click method (right).

    Automatic detection of hypoperfused areas in SPECT

    The fully automatic intersubject registration technique in CBA allows for reference images representing "normality" to be created by averaging images from helthy volunteers. Images from a patient can then, after a preceeding normalization of anatomy and intensity, be subtracted from the reference volume. Deviations in the patient data from the normal state can be detected in this way and can be displayed in subtration images and t-test images.