Michelle Hampson, PhD: The use of real-time fMRI to treat and study mental illness

December 01, 2020

Information

Associate Professor of Radiology and Biomedical Imaging; Director of real-time fMRI

ID5935

To CiteDCA Citation Guide

00:00My name is Michelle Hampson.
00:01I'm going to be talking today about
00:04some work that our group has been doing
00:06using real time fMRI neurofeedback to
00:08try to treat and study mental illness.
00:10So what do we mean by real time?
00:12fMRI neurofeedback?
00:13Well, this is a picture of somebody
00:15receiving this form of neurofeedback.
00:16They're lying in an MRI scanner
00:18and we're monitoring their brain
00:20activity patterns and we give them
00:21feedback over the course of the scan,
00:23showing them how some aspect of
00:25their brain activity is changing
00:27over the course of the of the scan
00:29and they get lots of practice trying
00:31to control that brain activity.
00:32Using the feedback that we give
00:34them as a training signal.
00:36So there's many different applications
00:38using this technique across the world
00:40to try to train people to control
00:42specific aspects of their brain function.
00:44I'm going to talk today about a
00:46specific application where we use
00:48this to try to train people to
00:49control contamination anxiety.
00:51So why am I focused on contamination anxiety?
00:54Well,
00:54it's a major symptom dimension
00:55of obsessive compulsive disorder,
00:57and it's also common in the healthy
00:59population.
00:59An regardless of whether you're a piece
01:02of patient with OC D or a healthy person.
01:05If you're experiencing contamination anxiety,
01:06we know that you have hyperactivity.
01:08And your orbital frontal cortex.
01:10So this is an area right at the front
01:12of your brain which provides kind of
01:14a biomarker of contamination anxiety.
01:16And we were wondering if we train people
01:19to control activity in this region,
01:21will that help them to better
01:23control their contamination anxiety?
01:24So that's the study I'm going
01:26to describe to you today.
01:27In this study we collected three
01:29different outcome measures.
01:30First of all,
01:31we assessed how well the training enabled
01:33people to control their target brain area.
01:35We also monitor brain connectivity,
01:37pattern changes.
01:38So we collected resting state data before
01:40the intervention after the intervention.
01:42To see if the training basically
01:44rewired their brain connectivity and
01:45then Thirdly and most importantly
01:47we had a measure of control over
01:49contamination anxiety that we collected
01:51before the training and after the
01:53training to see if the neurofeedback
01:54training improved people's ability to
01:56control their contamination anxiety.
01:58The last thing I'll mention before
01:59going on is that in addition to
02:02the subjects who got real neural
02:04feedback in this study,
02:05we had a control group that received
02:07a fake or sham form of neurofeedback.
02:10So just so you have a sense of what the.
02:13Neurofeedback is like for participants.
02:14I'm going to show you a video,
02:16which is exactly what one of our
02:18subjects saw in the neural feedback training.
02:20On the left is an arrow that cues
02:22the mess to their current task.
02:23When it's a white arrow pointing
02:25forward that tells them just to rest,
02:27relax and look at the neutral image
02:29at other periods of time it will be a
02:31red arrow pointing up or a blue arrow
02:33pointing down and that skewing them
02:34to try to increase or decrease
02:36activity in the target brain area.
02:37And during those periods of
02:39provocative images shown depicting
02:40some contamination related thing.
02:41So I've started playing the video
02:42and you can see the feedback is
02:44provided here in this line graph.
02:46At the bottom of the screen.
02:48And right now they're just
02:49resting and looking at what's
02:50happening in their orbital cortex,
02:51but now it's an increased block,
02:53so now they're trying to
02:54get the line to go up.
02:56They can look at this picture
02:57of a dirty stove and think,
02:59oh that's a food surface
03:00that really should be clean.
03:01Whatever is going to be helpful to
03:03activate this anxiety related brain
03:04area Now it's going to decrease block,
03:06so they're going to decrease.
03:07Block is to bring the activity
03:08in this area back down there
03:10supposed to look at the image,
03:11but they can be appraised it you
03:13know you're in is usually sterile.
03:15There's nothing to be worried about.
03:16Whatever they think is going to be helpful.
03:18For controlling that region and now
03:20it's arrest block so they basically
03:21cycle through these three blocks,
03:22getting lots and lots of practice
03:24trying to get activity in the target
03:26brain area to increase in the red
03:27blocks and decreasing the blue blocks.
03:29And the sham subjects.
03:32Our match one to one to the real
03:34neurofeedback subjects an are shown
03:35at the time courses from the real
03:37neural feedback subjects brains,
03:38but misled to believe that getting
03:39feedback from their own brain.
03:41So to the extent that the real
03:43neurofeedback subject got the
03:44line higher in the red than the
03:45blue blocks there Matcham subject
03:46is misled to believe they're
03:48having a similar level of success.
03:49So that was the kind of control that we used.
03:52If you want to know anything
03:53else about our methods,
03:54we have a job paper with a
03:56cheesy video you can watch,
03:57but I'm going to present to
03:59you now the results.
04:00So first of all,
04:01did the training help them better
04:02control their target region?
04:04We evaluated that ability to
04:05control the target region at
04:06four points throughout the study.
04:08The important thing to note is
04:09that at the end of the training,
04:11they had statistically significantly
04:13greater control than they did at
04:14the beginning of the training.
04:16So indeed,
04:16it seemed like the neurofeedback
04:17training did improve their ability
04:19to control their target brain area,
04:20which is encouraging.
04:21But the more important question is,
04:23does it actually translate into
04:24anything outside of the scanner so
04:26they actually get better at controlling
04:28their contamination anxiety?
04:28So for that,
04:29we turn to our behavioral measure
04:31that we collected before and
04:32after the intervention.
04:33Indeed,
04:33what we found.
04:34Is that the training did improve
04:36ability to control contamination
04:37anxiety in the neurofeedback
04:39subjects and it had no significant
04:41effect and the control group.
04:42So this was really exciting to
04:44us because it implies that this
04:45training can help people better
04:47control their contamination
04:48anxiety and contamination.
04:49Anxiety is a major symptom dimension of OC D,
04:52so that means that it has the same
04:54effects in the clinical population.
04:56This intervention would have
04:57direct clinical utility,
04:58but it raises the question well,
05:00what's going on in the brains of
05:02these people that they're better
05:04able to control their contamination
05:05anxiety after the training.
05:07So for that we looked at the resting
05:09state connectivity changes we used
05:11to measure of global connectivity,
05:12where we evaluate which parts of the
05:15brain changed their global amount
05:16of connectivity over the course of
05:18the training and what we found is
05:20that these areas shown with red,
05:22yellow colors,
05:23whereas it had significant
05:24increases in global connectivity
05:25in the neurofeedback subjects.
05:26These are prefrontal control areas
05:28and the areas shown in blue had
05:30decreases in global connectivity,
05:31so these are a lot of limbic
05:33emotion processing parts of
05:35the brain including insula,
05:36substantia nigra, amygdala, hippocampus.
05:37So it seemed like this is very
05:40consistent with neurobiological
05:41models of emotion regulation,
05:42whereby prefrontal control areas come
05:44online to help downregulate processing
05:46in limbic or emotion parts of the brain.
05:48And it seemed like the neural feedback
05:50was strengthening that circuitry an I
05:52should note that this is these effects
05:54were specific to the neurofeedback group.
05:57We did a similar analysis in the control
05:59group and there were no significant
06:01changes in global connectivity.
06:02But this is the pattern we see in the
06:05group of neurofeedback subjects as a whole.
06:08An within this group there are a lot
06:10of differences between subjects,
06:11so some subjects really get a lot from
06:13the intervention and other subjects find
06:15it frustrating and get very little from it.
06:17So we wanted to see what's happening in
06:19the brains of those good responders,
06:21and we computed correlation map
06:22examining which parts of the brain
06:24had changes in global connectivity,
06:25and the people who really responded well.
06:27And what we found jumping out
06:29of this analysis.
06:30This is analysis that treats every
06:31part of the brain in the same way,
06:33but what popped out of this
06:35analysis was our target region
06:36of the orbital frontal cortex.
06:38So the people who were rewiring
06:40this target area where the people
06:41who are really gaining control
06:43over their contamination anxiety,
06:44which is encouraging that we're
06:46targeting a reasonable part of the brain.
06:48If you manage to rewire the brain area,
06:50you get better control over your anxiety.
06:53So we thought all this data was
06:55really promising and we decided we
06:57wanted to move forward in studying
06:58this in an OC D population.
07:00So we applied for funding from NIH and
07:03we're now running a clinical trial in OC T.
07:05This is a randomized double blind
07:07placebo controlled trial which were
07:09actually close to wrapping up now.
07:11It's very similar to the study
07:13I just described to,
07:14except that we're running OC D
07:16patients and we're monitoring
07:17symptoms with classic symptom scales
07:18rather than a behavioral measure.
07:20But one interesting new aspect
07:21of the study is that
07:23we're following these subjects up clinically
07:25after they receive the neural feedback.
07:27So in the study I just described to you,
07:30we assess their ability,
07:31their, their contamination,
07:32symptoms of a few days after the training,
07:35and we do that in this
07:36clinical trial as well.
07:38But in addition, we bring them back
07:40two weeks later and assess them again.
07:42Then we also bring back a month
07:44later and assess their symptoms
07:45amonth after the training.
07:47So in this clinical trial we're
07:49actually following them from month
07:50post neurofeedback and the reason we're
07:52doing that as we wanted to rule out
07:54the possibility that the effects we
07:56are inducing were just transient so it
07:58wouldn't be super clinically useful.
08:00If you know participants who got
08:01real neurofeedback got better
08:02while they're receiving the neural
08:04feedback and their symptoms,
08:05you know decreased,
08:06and then as soon as you know feedback ended,
08:09they regressed back to baseline.
08:10That wouldn't be super clinically useful
08:12'cause you kind of need something
08:14that persists in a person's life.
08:16What we were hoping to show by doing
08:18this follow up is that the improvements
08:20in symptoms that we induced with the
08:22training were maintained during the
08:24month that we were following up afterwards,
08:26but we found was actually quite surprising.
08:28The subjects got better
08:29during the neurofeedback,
08:30but they continued to get better for
08:31a month after the neurofeedback so
08:33that the biggest differences between
08:35the control group and the real
08:37neurofeedback group were actually
08:38a month after training.
08:39So this was totally surprising
08:40to us and we thought,
08:42is this specific to OC D is
08:43this like an idiot?
08:45Syncratic phenomena for this application?
08:46But then we.
08:47We went and looked at another study
08:49we were running in Trett syndrome,
08:51which showed exactly the same pattern.
08:53This was a different population being
08:54trained on a different brain area,
08:56but we are assessing different
08:58type of symptoms.
08:59But we saw the same qualitative pattern,
09:01so we thought,
09:01well,
09:02maybe this is not an uncommon pattern
09:04in neurofeedback experiments and so we
09:05wrote this up and published in neuroimaging.
09:08It since heard from multiple groups
09:09that they are seeing similar
09:11patterns in other applications.
09:12So it's a really important for the people
09:15designing studies in our field to know this.
09:17Because if you're going to.
09:19Running her feedback study,
09:20it's very important to follow people
09:22up for a month or you might be missing
09:25the time point of greatest effect.
09:27So this that I've described here is just
09:29one of the studies were running at Yale.
09:31There are many different ongoing studies
09:33looking at all sorts of different disorders.
09:34If you're interested in using real
09:36time after a feedback to treat
09:38or study mental illness,
09:39please come talk to me.
09:42I want to just quickly acknowledge
09:44contributions of the people.
09:45This I can't possibly show everybody
09:47doing their feedback research at Yale.
09:49This is the subgroup that was involved
09:52specifically in that sub clinical OC D study,
09:54our primary clinical collaborator
09:56in that setting is Chris Pittenger.
09:58Teodora Stoica collected all that day.
10:00That I presented to you and Dustin
10:02Chinese has been really involved in,
10:04you know, creating and maintaining the
10:06real time imaging systems that we use.