Section: Application Domains
parietal aims at proposing innovative techniques to study brain function through the analysis of anatomical and functional brain images. Although much work has been performed in this field since the mid 90's, and standard solutions have been proposed - in particular a procedure called statistical parametric mapping (SPM), which has been progressively elaborated from 1995 to ca 2005 and implemented in several software packages- some important issues still need to be addressed:
First, quite surprisingly, standard frameworks do not fit with the way neuroscientists think of their data, that is the output of a highly modular and densely connected network, with a spatial layout that can be characterized through generic anatomical constraints.
Second, in spite of the intuitive evidence -most of the blood oxygen-level dependent (BOLD) activity seen in functional Magnetic Resonance Imaging (fMRI) originates mainly from the cortex- analysis on the cortical surface is not a standard yet. Spatial models are thus quite coarse and not informed by the anatomy.
Third, a crucial problem within the standard approach is that it still relies on poor approximations to deal with multiple comparison problems in statistical inference procedures. More work is needed in inter-subject modelling and reproducibility assessment.
Finally, it is still necessary to understand and characterize the informative content of neuroimaging activation maps, beyond the traditional maps of activity. In particular, the neuroimaging community should benefit from the current advances in machine learning and computational neuroscience.