Neuroimaging, big data and mental health: A survey of the land

by

This fig­ure shows how dif­fer­ent brain net­works func­tion togeth­er; the lines mea­sur­ing the strength of com­mu­ni­ca­tion among var­i­ous net­works. By using such dis­plays, we can eval­u­ate what is called the human func­tion­al connectome—and also how men­tal ill­ness is dis­rupt­ed. [Fig­ure cour­tesy of Vince Calhoun]

Men­tal and degen­er­a­tive dis­or­ders are among the most cost­ly and com­mon caus­es of dis­abil­i­ty in soci­ety today. Because the brain is the most com­plex organ in the human body, diag­nos­ing and treat­ing prob­lems when things go wrong pos­es enor­mous chal­lenges. Even before the 1990s was des­ig­nat­ed the Decade of the Brain, the poten­tial of neuroimaging—the tech­nol­o­gy that makes it pos­si­ble to see inside the work­ing brain—was a major focus in psy­chi­a­try. Since that time, expec­ta­tions have been high that neu­roimag­ing would move the nee­dle for­ward in unrav­el­ing the mys­tery of men­tal illness. 

Liv­ing in a time of such nation­al pro­grams as the BRAIN Ini­tia­tive and Human Con­nec­tome Project, we’ve become accus­tomed to hear­ing terms such as arti­fi­cial intel­li­gence, vir­tu­al real­i­ty, and brain-machine inter­face. In light of the expo­nen­tial increase in com­pu­ta­tion­al and algo­rith­mic pow­er, one can only assume that we have made great progress via psy­chi­atric neu­roimag­ing. But just how far have we actu­al­ly come in the last half-cen­tu­ry? How close are we to hav­ing neu­roimag­ing-based tools that can be used in the clin­ic? Have we learned any­thing about diag­no­sis or men­tal ill­ness from the vast trove of neu­roimag­ing data that has been col­lect­ed over the years? I wish I could point to spe­cif­ic exam­ples where the widen­ing use of neu­roimag­ing is begin­ning to help the men­tal­ly ill, but we are just not there yet. » Keep read­ing the excel­lent arti­cle The Promise of Big Data Imag­ing for Men­tal Health over at Dana Foun­da­tion’s Cere­brum.

News in Context: