Last week, I came across a paper entitled "The synchronous neural interactions test as a functional neuromarker for post-traumatic stress disorder (PTSD): a robust classification method based on the bootstrap". I've been mulling over whether to blog about this paper; on one hand, it presents some interesting conclusions, which I greet with some degree of skepticism. On the other hand, having training neither in human brain imaging, psychology, or post-traumatic stress disorder, I was loath to make any statements regarding the paper and it's conclusions.
However, after reading the paper, and discussing its existence with a number of people, I've decided to publish a post (this one) asking for the opinions of those readers who do know enough about the subject and the methodologies utilized to comment on the validity of the papers conclusions. Below I will describe the paper's methods, its conclusion, and a link to the article. Please, comment on the paper and its conclusions.
The paper, by Georgopoulous et al, published in the Journal of Neural Engineering, describes the use of magnetoencephalographic (MEG) recordings to diagnose PTSD. Recordings of synchronous neural interactions (SNI) were made while asking subjects to fixate on a spot of light for 60 seconds, allowing the researchers to "engage the brain in a stable condition". The recorded data were analyzed using "bootstrap-based analyses". The paper explains the choice to use a bootstrap-based analysis:
"This approach was motivated by the common practical situation where a ‘new’ subject S needs to be classified, given subjects of known condition, i.e. belonging to a control (i.e. non-PTSD) group C or to the PTSD group D."
Using their analysis, which is detailed in the paper, the authors claim they can successfully differentiate subjects suffering from PTSD from healthy control subjects with greater than 90% accuracy.
Conclude the authors:
"Altogether, these findings document robust differences in brain function between the PTSD and control groups that can be used for differential diagnosis and which possess the potential for assessing and monitoring disease progression and effects of therapy."
A note: although all PTSD subjects were veterans, not all were suffering from PTSD induced by military trauma. Some were affected due to childhood trauma or non-military trauma during adulthood. Given the differences in the causation, is it reasonable to expect that all suffers of the disorder would display distinguishable synchronous brain activity differences? Certainly, I do not find it impossible that the data could show differences in brain activity when comparing PTSD sufferers from healthy controls. But in my mind, a diagnostic tool is most useful if it can differentiate one psychological issue from another. Is it reasonable to expect that a patient will either have PTSD, or be completely psychologically normal? I wonder about the selectivity of the SNI test. Can it differentiate between PTSD and any one of the myriad of psychological issues that can affect the human brain (especially following military service). The ability to distinguish between shades of grey, instead of black versus white, must be the final goal of any psychological diagnostic tool.
As previously mentioned, due to my inexperience with brain imaging and mathematical techniques of analyzing such data, I cannot comment of the use of the bootstrap analysis. Is there someone out there willing to read the paper and provide an analysis?
Georgopoulous et al. The synchronous neural interactions test as a functional neuromarker for post-traumatic stress disorder (PTSD): a robust classification method based on the bootstrap. J. Neural Eng. 7 (2010). doi:10.1088/1741-2560/7/1/016011
Updated 1/25/2010: Have just seen that Vaughan Bell over at Mind Hacks wrote on the 24th about this article, brining up that same issue regarding the scan's ability to separate PTSD from normal, as opposed to PTSD from another disorder.