Specification and Selection of Regions of Interest (ROIs) in a Computerized Brain Atlas

Abstract

The computerized individually adjustable brain atlas (CBA) has been further developed. The atlas was primarily designed for anatomical localization and quantitative evaluation of data in positron emission tomography (PET), but may also be employed for other neuroimaging modalities, such as transmission computed tomography (CT) and magnetic resonance imaging (MRI). The atlas is based on anatomical information obtained from digitized cryosectioned brains. Using spatially standardized and then averaged MRI images, we demonstrate the high localization accuracy and precision of the brain atlas. This is a prerequisite for obtaining accuracy when using the atlas in the localization and the quantitative evaluation of PET data. The specification and the selection of region of interests (ROIs) by the CBA are presented and discussed.

Accuracy and Precision of the Computerized Brain Atlas Programme for Localization and Quantification in Positron Emission Tomography

“…More recently, this computerized brain atlas programme (CBA) has been further developed. It now enables one to spa tially standardize (reformat) individual image data to provide precise interindividual comparisons of functional PET image data (Bohm et al, 1986(Bohm et al, , 1990 Greitz et al, 1989).…”

“…More recently, this computerized brain atlas programme (CBA) has been further developed. It now enables one to spa tially standardize (reformat) individual image data to provide precise interindividual comparisons of functional PET image data (Bohm et al, 1986(Bohm et al, , 1990 Greitz et al, 1989).The anatomical data base of the CBA was ob tained from digitized cryosectioned human brains and includes the brain surface , the ventricular sys tem, and �250 structures. The anatomical nomen clature of the cerebral cortex and brainstem struc tures is the revised Nomina Anatomica of the In ternational Nomenclature Committee (1968).…”

A methodology for specifying PET VOIs using multimodality techniques

“…Region placement approaches can be classified roughly into one of two groups: anatomy based and physiology based. Anatomy-based approaches, like the one described in this paper, rely on either coregistered computed tomography (CT) or MRI data on which regions are placed, or on standardized region templates which are typically stretched or warped in some manner to best fit the current PET data [24]. Physiology-based approaches rely only on the PET data, typically drawing regions around or automatically detecting areas of peak activity via subtraction or statistical techniques on parametric images which can be derived without anatomical information [25], [26].…”

The use of a computerized brain atlas to support knowledge-based training in radiology

“…If successful, the structures outlined and labelled on the reference illustration will be accurately superimposed on the corresponding ones from the case image. The same technique can be used to register images of low spatial resolution, such as PET, SPECT and MEG, with high resolution CT or MRI reference images to locate functional activities on the brain (Fox, Perlmutter et al 1985;Friston, Passingham et al 1989;Seitz, Bohm et al 1990;Bohm, Greitz et al 1991;Evans, Marrett et al 1991;Barillot, Lemoine et al 1993;. Image interpretation can also be carried out symbolically, using model-based image interpretation and symbolic inferencing techniques to identify segmented regions on a case image (Natarajan, Cawley et al 1991).…”