Sports-related concussion differentially impacts functional brain networks in college athletes
Sports-Related concussion differentially impacts functional brain networks in college athletes
Toggle between list and paragraph view.
- [00:00:01.230]my name is Zach Headley and I will
be going over my UCARE project today,
- [00:00:04.890]titled "Sports related concussions,
- [00:00:07.980]functional brain networks in
college athletes." Athletes receive
- [00:00:13.530]around 2 to 4 million sports
related concussions each year in the
- [00:00:17.310]United States. Sports related
- [00:00:19.290]concussions are complex brain
injuries causing multiple changes in the
- [00:00:23.340]brain, including the disruption of
functional brain networks. Overall,
- [00:00:27.450]this project uses magnetic resonance
imaging to identify changes in individual
- [00:00:31.950]network organization
due to sports related concussion.
- [00:00:35.900]In order to complete this study.
- [00:00:37.880]baseline MRI data was collected for
all incoming UNL football players
- [00:00:42.140]since 2018.
- [00:00:43.700]The present results will focus on data
from 26 football players diagnosed with
- [00:00:48.050]sports related
concussion. 24 to 48 hours
- [00:00:51.620]after receiving a sports
- [00:00:54.170]a player receives a post scan and
additional scan is taken right before the
- [00:00:58.910]player returned to play.
- [00:01:00.980]These scans showed the BOLD signals for
very small areas of the brain called
- [00:01:06.350]These voxels can be grouped into
264 larger regions called nodes,
- [00:01:10.910]which can then be grouped into
networks. Overall, we see,
- [00:01:15.110]we wanted to see how the network
assignments of nodes varied from baseline
- [00:01:20.000]to post-injury to return to play.
And in order to accomplish this,
- [00:01:24.260]we first had to assign
networks to each node.
- [00:01:27.200]The BOLD time series data of one node
was compared to the meantime series
- [00:01:32.090]of, for each network in the power Atlas.
- [00:01:36.140]And it was assigned to the network that
it was most closely correlated with.
- [00:01:40.550]All of the nodes were assigned to the
networks in this manner. After each node
- [00:01:45.260]had been assigned to a network,
- [00:01:46.910]we repeated this process
after recalculating the
mean network time series,
- [00:01:51.890]based on these new network assignments.
- [00:01:54.470]This process was iterated until 95% of
the nodes did not change their network
- [00:01:59.000]assignments from one
iteration to the next.
- [00:02:01.910]Once we found the stable network
assignments for all individuals,
- [00:02:05.750]for each of the three scans,
- [00:02:07.520]we examined the changes in network
assignments related to sports related
- [00:02:13.160]A key point to our research is
the collection of baseline data
- [00:02:18.530]that is needed to compare
separate individual posts, injury,
- [00:02:21.650]and return scared scanned accurately
due to the high variability of network
- [00:02:26.150]assignments to various
nodes across individuals.
- [00:02:29.450]The high variability of
network organization at
baseline is shown in this upper
- [00:02:34.010]middle matrix. This matrix shows
the functional connectivity of the,
- [00:02:38.220]of the 264 nodes across participants.
- [00:02:42.050]That color scale from black to yellow
represents the standard deviation
- [00:02:47.000]from the mean of the functional connectivity
with the yellow showing the highest.
- [00:02:51.680]variability. As you can
- [00:02:54.380]see in this matrix, the connections
across individuals differ considerably,
- [00:02:59.020]which is why it's important to collect
baseline data for each person. Next,
- [00:03:03.880]I'm going to show you the network
assignments we came up for with at
- [00:03:09.760]You can see these results on the
two brains depicted on the left. In
- [00:03:14.230]comparison to the power Atlas
organization of networks,
- [00:03:20.140]our network assignments are more random
and do not follow the general network
- [00:03:24.280]topography we'd expect to see.
- [00:03:26.890]There are some networks usually
located in specific locations,
- [00:03:31.690]such as the visual network in the
occipital lobe and the default mode
- [00:03:36.520]located in the posterior cingulate and the
medial prefrontal cortex. However,
- [00:03:41.260]with our original approach,
- [00:03:42.310]we saw a little correspondence
with these established networks,
- [00:03:45.760]instead observing that our networks were
randomly spread out through the brain.
- [00:03:52.020]After this observation, we
reevaluated our approach.
- [00:03:57.060]There are three
- [00:03:57.640]networks from the Power Atlas that
could be adding high variability to our
- [00:04:01.230]results due to their undefined nature:
- [00:04:03.930]the Uncertain, Sub-cortical
and Cerebellum networks.
- [00:04:07.620]Also the high threshold could be
causing additional variability,
- [00:04:10.980]forcing the nodes to be reassigned more
than necessary. Based on this evidence,
- [00:04:15.600]we modified our approach
to focus on 11 networks,
- [00:04:19.950]excluding those previously stated,
and use a lower threshold of 50%.
- [00:04:25.230]This approach created the network
organization shown on the right in
- [00:04:30.000]comparison to the Power Atlas.
- [00:04:32.730]These two right
- [00:04:33.810]brains are more consistent with
the known network organization.
- [00:04:37.980]For example, the visual network,
- [00:04:40.380]shown in blue, is located
- [00:04:44.340]in the occipital lobe is more apparent
with this approach compared to the
- [00:04:50.220]The default mode network
is more consistent with the
- [00:04:54.420]on this network than
- [00:04:57.180]the new approach provided
network assignments that
were more consistent with the
- [00:05:01.740]known network architecture
of the brain. Next,
- [00:05:05.130]we compared the number of nodes per
network at baseline compared to post these
- [00:05:09.900]results are shown in the upper right
- [00:05:12.540]hand corner. We did not find
- [00:05:15.630]any significant differences in the
number of nodes assigned to each network
- [00:05:19.440]following concussion. Some networks
had more nodes at baseline,
- [00:05:23.490]while other networks had
more nodes after concussion.
- [00:05:28.380]We are currently
- [00:05:29.220]analyzing the possible recovery of nodes
to their baseline network assignments
- [00:05:33.840]before the football player
begins to play again.
- [00:05:37.140]We specifically focused on hubs because
they are crucial to the functional
- [00:05:42.270]in the brain. Hubs are critical
- [00:05:44.310]nodes that interact with many more
nodes in the network and across networks
- [00:05:48.480]compared to other nodes. Participation.
- [00:05:51.240]coefficient is a specific statistic that
measures a node's hub characteristic.
- [00:05:56.250]We can also identify hubs in the air traffic
patterns across the United States.
- [00:06:01.310]An airport like Atlanta would be a hub
with a large participation coefficient
- [00:06:05.840]because is connected to
many different airports.
- [00:06:09.080]While an airport like Lincoln would not
be a hub and have a small participation
- [00:06:12.860]coefficient. We'd looked at the
proportion of nodes after concussion,
- [00:06:17.390]and that returned to play that were
assigned to the same network they were
- [00:06:22.070]at baseline. We focused
- [00:06:24.590]on the networks with the largest
- [00:06:27.650]which were Default Mode, Memory
and Dorsal Attention networks.
- [00:06:31.340]These averages can be seen
in the right middle square.
- [00:06:36.410]At return
- [00:06:36.920]to play,
- [00:06:37.580]there was a greater proportion of
nodes that recovered to their baseline
- [00:06:41.300]networks. However,
- [00:06:42.320]this was not statistically significant.
Future analyses will continue to examine
- [00:06:47.120]how hubs may be impacted by
sports related concussions
- [00:06:52.880]To conclude, baseline data is crucial
- [00:06:55.810]To conclude,
- [00:07:00.910]baseline data is crucial to the high
variability and functional connectivity
- [00:07:05.200]across participants
Through a parameter search,
- [00:07:08.380]we found that a less stringent stability
threshold did the best job of capturing
- [00:07:12.100]individual network organization while
preserving the well-established general
- [00:07:16.120]network topography.
- [00:07:18.070]We did not find any significant changes
in the number of nodes assigned to each
- [00:07:22.090]functional network from baseline
to post-injury. Future analyses
- [00:07:25.930]will focus on the stability of networks
- [00:07:27.880]hubs compared to nodes that
are not classified as hubs.
- [00:07:33.010]I'd like to acknowledge my co-authors
for their assistance and contributions as
- [00:07:37.150]well as the NIH Great Plain
IDeA CTR Network Pilot Grant,
- [00:07:42.100]and UCARE for allowing me to
complete this project. Thank you.
Log in to post comments