Brain-wide association studies, which use magnetic resonance imaging to identify relationships between brain structure or function and human behavior or health, have faced criticism for producing results that often cannot be replicated by other researchers.
A new study published in Nature demonstrates that careful attention to study design can substantially improve the reliability of this type of research. For the study, Kaidi Kang , a biostatistics PhD student, Simon Vandekar , PhD, associate professor of Biostatistics, and colleagues analyzed data from more than 77,000 brain scans across 63 studies.
The investigators found that studies can achieve more dependable results by strategically selecting study participants to ensure a wider range of the characteristics being measured. For example, when studying age-related brain changes, including more participants at both younger and older age ranges produces more reliable findings than random sampling or concentrating on middle-aged participants.
The research also revealed that collecting multiple brain scans from the same person over time can improve reliability for some brain measurements but may actually reduce reliability for others, depending on what is being studied.
Based on their findings the authors offer practical guidance for scientists designing future brain imaging studies, potentially helping them obtain reliable results with fewer participants.
Kang and Vandekar were joined by investigators from six institutions in the U.S. and the United Kingdom. Others on the study from Vanderbilt are Jiangmei Xiong, MS, Megan Jones, Ran Tao, PhD, and Jonathan Schildcrout, PhD. The study was supported by the National Institutes of Health (grants R01MH123563, R01MH132934, R01MH133843, K23DA057486, R01MH120482, R01MH112847, R01MH113550, R01EB022573, R37MH125829).
