Magnetic Resonance Imaging (MRI) is an important research area, and it’s also a “hot topic” with periodic excited claims in the media. I discuss this in the context of various new treatment claims for ADHD and try to separate the wheat from the chaff in my book, Getting Ahead of ADHD. So far, the bottom line is that brain imaging can’t diagnose ADHD or its subtypes, despite frequent claims that it can (based on single, small studies).
My research team and others are now undertaking a new generation of studies using advanced nonlinear equations (called machine learning) to improve prediction. These are powerful methods and there is reason to be hopeful. However, results from existing studies do not yet offer clinical value. One limitation is that sample sizes tend to be extremely small (often less than 100 children)—such samples are prone to chance findings that will not generalize, not matter how clever the analysis is. Scientists attempt to overcome this with re-sampling methods within their study (one common method is called “k-folds” cross validation). That is of some help, however, the acid test is whether the prediction holds up in a completely new, independent sample of children. Typically, that check is not even done. When it has been done, results are generally disappointing. This generalizability problem is challenging and it will be some time before it is solved. However, many research groups, our own included, are trying hard to do this. I have high hopes that we will eventually succeed. So stay tuned, but for now, remain skeptical of claims for breakthrough brain imaging diagnostics for ADHD. I do not advise you to seek brain imaging for a clinical case of ADHD in the absence of other medical indication and nothing has changed with recent publications or press reports. Check back here or sign up for e-mail updates. We will be tracking this literature and I will share significant findings with you.
As always, let’s keep our eye on the science for reliable answers.