A potential game-changer has emerged in the world of brain network mapping, raising questions about the reliability of a widely used method. Over 200 published studies and ongoing clinical trials may have relied on faulty brain network maps, according to a recent study published in Nature Neuroscience.
The method, known as Lesion Network Mapping (LNM), has been a popular tool for identifying patterns of brain activity associated with various neurological conditions. However, the new study suggests that a mathematical flaw in LNM leads to nearly identical network maps for different datasets, casting doubt on the method's accuracy.
The Impact of a Methodological Flaw
LNM has been used to map networks for conditions like obsessive-compulsive disorder, Parkinson's disease, and psychopathy, offering a promising way to target specific brain networks for treatment. But the new study reveals that these networks are not rooted in the biology of the condition, but rather, are a result of the method's mathematical error.
This means that the targets identified by LNM are unreliable, as they may be correct by chance but often lead researchers astray.
A Controversial Finding
But here's where it gets controversial: the creators of LNM argue that the issues raised are not insurmountable. They acknowledge that the method can produce false-positive or nonspecific findings, but believe that careful study design can account for these limitations.
And this is the part most people miss: the problem lies in how the datasets are processed. Each new network added to a lesion network map primarily derives from the original typical connectome, leading to maps that increasingly resemble the basic properties of that connectome, with minimal disorder-specific contributions.
Moving Forward
So, what now? Some researchers suggest starting fresh with a new form of network analysis, as the problem stems from the data analysis itself. Others believe that there may still be useful disease-specific information within LNM maps, and efforts are being made to address the known issue of LNMs converging to the typical connectome.
Despite these potential solutions, the controversy surrounding LNM highlights the need for caution when translating new platforms and pipelines to clinical use. It also encourages researchers to explore alternative methods for mapping brain networks involved in neurological disorders.
What do you think? Should we continue to use LNM with careful study design, or is it time to move on to new methods? The floor is open for discussion.
