Dubois Bowman, Biostatistics and Neuroimaging

 


Can you imagine mapping the mind in action?  DuBois Bowman, Associate Professor of Biostatistics, is doing just that.  He develops statistical methods for analyzing brain functioning from an expanded spatial-temporal vantage point.   

The human brain can be represented by hundreds of thousands of tiny volume elements or "voxels" (a few cubic millimeters in size), with each containing a localized measurement of brain activity.  The activity in each voxel is constantly changing and voxels associate with one another in intricate ways.  Their association occurs both within one region of the brain (intra-regional coherence) and across regions (inter-regional connectivity).  Typical analyses of neuroimaging data treat each voxel as separate or discrete, therein depicting mental processes as a series of isolated actions. But scientists long have known that the brain is more complicated.  Bowman creates statistical models that more accurately capture the varying activity of these voxels and their intricate dynamism. "My models build an integrated network between brain regions, so that we can now pool information from related voxels," he says.  His scholarship holds special promise for applications in clinical settings.

In particular, Bowman has developed neuroimaging statistical methods that deepen our understanding of brain activity within cocaine-dependent individuals, related to their impaired ability to inhibit a prepotent response.  Using a spatial map of the brain with 90 pre-specified anatomical regions, he has made it possible for scientists to measure differences in brain activity between cocaine-dependent individuals and a group of healthy controls and to examine neural processing changes due to treatment.  Preliminary findings from the study suggest that cocaine-dependent people develop "inhibitory control-related deficiencies in neural processing," Bowman says.  In response to treatment, the cocaine-dependent brain appears to devise coping strategies to circumvent these shortcomings that rely on the use of more interactions between brain regions. The brain develops its own plan for recruiting healthy regions to help supply areas with deficiencies.

As scientists study the brain's capacity to design personalized compensatory strategies, they are recognizing that effective treatment demands a similarly personalized approach. Bowman's scholarship also includes predictive modeling: developing statistical algorithms that will allow clinicians to predict how a mental health patient will respond to different treatment options.  His current algorithms target the treatment of persons with schizophrenia. Says Bowman: "Doctors will be able to assess the merits of one treatment intervention versus another by combining population-level information with personalized neuroimaging characteristics and other factors such as genetic and demographic information."  Though predictive neuroimaging will not replace the professional judgment of physicians working over time with their patients, they promise a useful resource for testing treatment options.

On a broader level, Bowman's scholarship also invites consideration of how and why the human brain "reasons" – in particular, how it undertakes the task of moral reasoning.  He recently collaborated with an Emory research team led by faculty members Diana Robertson in the Goizueta Business School and Clinton Kilts in the School of Medicine to study the neural foundations of moral awareness in relation to care and justice issues.  The study, which assessed the brain activity of students in Emory's Executive MBA program, found that certain regions of the brain are ignited or activated by tasks that entail "moral" content.  What is more, the brain responds differently depending on the precise nature of the morally sensitive task: one part of the brain fires up in response to a "justice" issue, another to "care" issues.

For Bowman, who joined the faculty in 2000, working on such interdisciplinary projects is one of the best things about being at Emory.  Within his department, he has established a Neuroimaging Biostatistics Research Group (NBRG).  Outside his department, he collaborates with faculty in Psychiatry, Radiology, Anesthesiology, Business, Biomedical Engineering, and elsewhere to apply his statistical models to a variety of imaging studies. "Emory builds bridges between silos," Bowman says.  "That's essential for doing this work.  We can advance knowledge more as a team than we can by trying to tackle these problems separately."

Maybe "intra-regional coherence" and "inter-regional connectivity" describe not only those enterprising voxels but also the thriving university.