We have presented some of the recent developments in the field of computational models for brain image analysis. Despite their many current limitations, modern computational models for brain image analysis have made it possible to examine brain structure and function in greater detail than was possible via traditionally used methods. For example, such models will soon allow us to obtain structural and functional measurements regarding individual cortical gyri or sulci, not merely of larger structures such as lobar partitions of the brain. Moreover, we will be able to perform such measurements on large numbers of images, because of the high degree of automation of algorithms emerging in this field.

Despite the recent progress, several issues will need to be addressed in the future. In particular, most investigators have focused on analysis methods for the normal brain, or for brain that has been affected in relatively subtle ways by a disease. Hence, it has been possible to map brain images from one individual to those of another via transformations that are one-to-one and onto. However, in many cases, gross morphological changes occur in the brain, such as in the development of tumors. Models that deal with such cases are still in their infancy [41,42]. As a second example, we note the analysis of images from animals whose genetic composition is altered, so that morphological and physiological effects can be measured [43]. Such genetic mutations can cause abnormalities well beyond the ones that can be handled by current models and algorithms.

Computational neuroanatomical models will evolve in various directions during the next decade. However, the main foundation that allows investigators to examine brain structure and function in precise, quantitative ways has already been laid.

0 0

Post a comment