Altered Global Brain Signal in Schizophrenia

Summary of my PNAS 2014 paper of the same title.

I published my first first-author paper in 2014 on a neuroimaging feature called the ‘global’ signal, the spatial average of all blood-oxygen level-dependent signals emitting* from the brain’s gray matter. This simple average is typically discarded in neuroimaging studies due to methodological reasons, and treated as clinically meaningless. However, I showed that this signal is profoundly altered in schizophrenia. Here’s a Yale article on it: http://medicine.yale.edu/news/article.aspx?id=7379  I’ll also summarize below, but first I’ll go over some basics..

*More specifically, when undergoing functional magnetic resonance imaging (fMRI), the magnetic field(s) generated during the scan have differential impact on the blood in the brain, depending on the oxygenation level of the blood. With the right magnetic resonance acquisition parameters, we can obtain what is called the blood-oxygen level-dependent (BOLD) signal.

Since the blood-oxygenation level is impacted by local metabolic activity of the brain cells, observing blood-oxygenation patterns over time (via magnetic resonance imaging) can give us an idea of the brain activity patterns in the person being scanned. Typically, the spatial resolution of the scanner allows us to obtain BOLD signal from 3-6 cubic mm-size volumetric pixels (voxels) in the brain. For reference, at 1x1x1mm final voxel size, the number of voxels in the brain is approximately 1,574,376.

The global signal in each person’s scan is computed by averaging across the time-varying BOLD signals of all voxels in the gray matter of that person’s brain.

————————————————————————————————————————–

Summary of my paper:

In a nutshell, it is possible to measure how much a BOLD signal fluctuates over time, by taking the variance (standard deviation squared) of the signal. You can repeat this for every BOLD signal coming from every location (voxel) in the gray matter (as opposed to white matter, or ventricles) portion of the brain. We specifically did this for gray matter in the cerebral cortex. I found that the variance of the BOLD signal in voxels from the gray matter portion of the cerebral cortex was increased (on average) in schizophrenia patients compared to healthy subjects. This was true with or without the global signal discarded. However, the difference between patients and healthy subjects was significantly reduced when global signal was discarded. In other words, patients were harder to distinguish from healthy subjects, if you removed the global signal. This potentially suggests that clinically meaningful (and perhaps diagnostically useful) information is contained in the global signal, and this may be part of the ‘signature’ of schizophrenia on brain activity patterns.