I think one challenge with bipolar is that the patients, themselves, are sometimes difficult to study in the kinds of studies that you'd like to really include them in. And so for example, individuals with bipolar disorder have difficulty with insight into their own illness. They often don't think they're ill, or they often don't think they need medicines.

And so, for example, individuals who are doing treatment trials or longitudinal treatment studies where if you're an imager and you're trying to do longitudinal studies, it's very difficult to keep patients in your treatment trial or in your longitudinal study.

Another challenge is that when patients are in the manic phase of bipolar disorder, then they often lose the ability to cooperate, and they often even lose the ability to provide informed consent, the capacity to provide informed consent.

And so we've learned a lot more about depression in the context of bipolar disorder than we have about mania in the context of bipolar disorder. But sometimes when you're following depressed patients and they become manic, then often they end up dropping out because they need to be hospitalized or they need to be placed on additional medications that then causes them to drop out of a longitudinal strategy.

I think one way is to be innovative in terms of - how are you flexible in ebb and flow with the course of people's illness? So for example, one of the ways we often get brain imaging studies conducted in various phases is we try to have longitudinal relationships with patients, and then they often take themselves off their medications.

And so when they're depressed and they have stopped their medicines, they're often willing to come back in for a study, and so we've been able to then get imaging accomplished during a depressed phase before we get them back on medications. And so that has been one important strategy for us.

I think another has been when you use non-pharmacological interventions like some of the psychotherapeutic techniques or some of the social rhythms techniques. My impression has been, and there is some gathering literature that this is true, it appears that by adding those kinds of non-pharmacological additions, then you actually improve the treatment compliance for the whole treatment strategy. So not only for those sort of non-pharmacological treatments, but also they're more likely to adhere to medicine treatment.