Connectivity in fMRI: Blind Spots and Breakthroughs
V. Solo, J.B. Poline, M. Lindquist, S. Simpson, F. Bowman, M. Chung
IEEE Trans Med Imaging, vol. 37, issue 7, July 2018
There is intense interest in brain connectivity analysis across all modalities due e.g. to its potential to deliver biomarkers for early diagnosis and to be an enabler of personalized medicine. But the realization of these potentials is being held up by blind spots and bottlenecks in the literature. How to compare brain networks with different numbers of nodes built on different time and spatial scales? How to make inferential methods out of the compelling exploratory tools of time-varying connectivity and graph analysis? How to use the auto-correlated information hidden in fMRI time series to realize personalized classification? This is a non-standard review paper which while noting current limitations, sketches compelling new methods from applied mathematics, mathematical sociology, statistics, and control engineering that promise to break through these blind spots.