Lenticular Nucleus

FIGURE 4 The figure on the right represents the average gray matter distribution for 20 individuals. These images have been spatially normalized to the template (atlas) shown on the left. Once the atlas and the image data are in a common stereotaxic space, labels can be transferred from the atlas to the data. Here we illustrate the process for two subcortical structures: the caudate and lenticular nuclei. Volumes can also be computed because the normalization procedure preserves the tissue volumetric units in stereotaxic space. See also Plate 15.

accurately deal with abnormalities such as tumors, atrophy, and anomalous tissue types.

In the functional brain image analysis field, segmentation allows information obtained about brain function, via functional MR imaging (fMRI) or positron emission tomography (PET), to be directly correlated with the underlying anatomy. Studies examining the association of functional activity with tissue volume, as in functional studies investigating and quantifying the effects brain atrophy in Alzheimer's patients, or research examining the functional importance of the cortical geometry, are made possible with accurate brain segmentation methods. In addition, improved localization, visualization, and quantification of brain compartments, yielded by image segmentation methods, are critically important to the fields of computer-aided neurosurgery, radiation therapy, surgical planning, drug delivery evaluation, and assessment of therapy effectiveness.

Many other challenges remain in the biomedical image segmentation field. Although most of our examples were applicable to MR images of the human brain, they are extensible to other organs, other species, and other imaging modalities. These techniques can make major contributions to the interpretation of images from the heart, lungs, abdomen, pelvis, and vascular system. They can also help in the quantitative analysis of vertebrate and microscopic imaging.

With increased utilization of new and improved biomedical imaging hardware in the hospitals, clinics, and research labs, and concomitant increase in image resolution and size, image segmentation stands as one of the most important and promising areas of image processing for biomedicine.

0 0

Post a comment