Extended Talairach Landmarks on Neuroimages for Atlas Registration

Abstract

The accuracy of scan-to-atlas registration highly depends on the number of landmarks and the precision of landmark identification. An extended landmark, cerebellum inferior (CBI), is introduced in this paper. The extracted brain and midsagittal plane are applied to identify the modified Talairach landmarks and the new introduced landmark CBI. The AC-PC plane is firstly determined and then anatomical information is applied to estimate the other landmarks. The proposed method is fully automatic and has been validated on 49 FDG-PET normal and pathological scans qualitatively, and 15 cases quantitatively. The average processing time is about 3 seconds on a standard personal computer.

Rapid and automatic atlas-based approach of Alzheimer's disease assessment by positron emission tomography neuroimages

“…A set of modified Talairach landmarks, conceptually equivalent to the original Talairach landmarks, was introduced [20] to overcome the problems and limitations of the original landmarks and become more constructive and easier identification by computer program. It is extended to include the cerebellum of a brain into the Talairach space as well [21].…”

“…The spatial normalization plays the role to process the PET images and transform them into the atlas space. It consists of three steps: 1) the brain areas are segmented from the PET images by a threshold which is selected by a histogram graph automatically [26], 2) the set of landmarks is accurately defined in the atlas space and automatically detected on PET images [21], and 3) based on the landmarks, the brain is divided into 18 cubic regions and a piecewise linear transformation is applied to generate a new volumetric image data fitting into the atlas space. Fig.…”