Brain cognition gaps reveal associations with dopamine and factors related to brain health through artificial intelligence prediction of functional connectome
- Department of Electrical Engineering and Computer Science, University of Stavanger, Norway
- Department of Diagnostic Imaging, Akershus University Hospital, Norway
- Institute of Clinical Medicine, University of Oslo, Norway
- Wallenberg Centre for Molecular Medicine (WCMM), Umeå University, Sweden
- Department of Medical and Translational Biology, Umeå University, Sweden
- Aging Research Center, Karolinska Institute and Stockholm University, Sweden
- Department of Diagnostics and Intervention, Diagnostic Radiology, Umeå University, Sweden
- Umeå Center for Functional Brain Imaging (UFBI), Umeå University, Sweden
- Department of Psychology, Florida State University, United States
Figures
Figure 1 with 1 supplement
Prediction of episodic memory from resting-state and task-based functional connectivity maps.
Model trained on functional connectivity maps acquired at rest predicts (a) episodic memory (EM) of the test dataset. The models trained on movie-watching (b) but not n-back (c) datasets predicted EM scores of the test dataset. Table 1 summarizes the p values, correlations, mean square error (MSE), and mean absolute error (MAE) for each model. Test datasets were obtained from the same cohort for rest, movie-watching, and n-back. The winning model trained on EM at rest was evaluated on the external COBRA cohort and yielded a significant prediction of EM scores (d). Bootstrap distributions of correlations between predicted and actual EM scores indicated no significant difference in the predictive power of EM between models trained on resting state and movie-watching data (e). Additionally, the bootstrap distribution revealed that models trained on resting state and movie-watching data yielded higher correlations than those trained on n-back data (e). Visualization of features contributing to the successful prediction of EM at rest (f). A Grad-CAM-derived saliency map displays the features that contributed to the model’s predictions. The hot spots overlaid on the FC map demonstrate noticeable cross-correlation contributions in ‘default mode’ (DMN) regions. Another important feature visualized by Grad-CAM includes off-diagonal hot spots reflecting inter-connections of the DMN–‘subcortical’ node.
Figure 1—figure supplement 1
Replication of EM prediction from resting-state and task-based functional connectivity using Schaefer-300 parcellation.
Figure 2 with 1 supplement
Prediction of working memory from resting-state and task-based functional connectivity maps.
Model trained on FC maps acquired at rest did not significantly predict (a) the working memory (WM) in the test dataset. The model trained on the movie-watching dataset yielded the best-performing model in predicting WM (b) while the model trained on the n-back dataset (c) was the second-best model. Table 2 summarizes the p-values, correlation power, MSE, and MAE for each model. (d) Results of cross-dataset validation, where the best-performing model in the DyNAMiC dataset (i.e., movie-watching) was applied to predict WM to the COBRA dataset. However, since COBRA does not include a movie-watching paradigm, we applied the model to the n-back task in COBRA (Table 2). Bootstrap distributions of correlations between predicted and actual WM scores showed no significant difference in predictive power between models trained on movie-watching and n-back data (e). The bootstrap distribution revealed that models trained on movie-watching and n-back data exhibited higher correlations than those trained on resting state data (e). The Grad-CAM-derived saliency map highlights dominant features in the FC maps that contributed to the model’s predictions (f). The hot spots overlaid on an FC map demonstrate noticeable cross-correlation contributions in the ‘VAN’, ‘visual’, and to a lesser degree (<0.5) ‘DMN’. Other important features visualized by Grad-CAM include off-diagonal hot spots reflecting inter-connections of the ‘DMN’ – with ‘FPN, fronto-parietal task control’, ‘Subcortical’, and ‘Cerebellar’; ‘Cerebellar’ – ‘FPN’ node.
Figure 2—figure supplement 1
Replication of WM prediction from resting-state and task-based functional connectivity using Schaefer-300 parcellation.
Figure 3 with 1 supplement
The plots compare physical activity scores (total hours per week), CVD risk, between two groups with positive and negative BAG as well as high and low memory performance in the DyNAMiC (top row) and COBRA (bottom row) datasets.
*, **, and *** denote p<0.05, p<0.01, and p<0.001, respectively.
Figure 3—figure supplement 1
Overview of cognitive tests included in the DyNAMiC study.
Figure 4 with 1 supplement
Relationship between the gap measured from predicted and actual EM scores and dopamine D1 and D2 receptors.
Relationship between the gap measured from predicted and actual EM scores and dopamine D1 receptor (a). Negative gaps indicate that the predicted EM score was lower than actual EM scores, while the positive gaps indicate more higher predicted scores than actual EM scores. Partial correlation analysis showed a significant correlation between D1 receptor values and the measured negative and positive gaps. Relationship between the gap measured from predicted and actual EM scores and dopamine D2 receptor (b). Negative gaps indicate that the predicted EM score was lower than actual EM scores, while positive gaps indicate more higher predicted scores than actual EM scores. Partial correlation analysis showed a significant link of D2 receptor values to negative gaps and positive gaps.
Figure 4—figure supplement 1
Correlations of brain age gap and brain-cognition gap with physical activity and cardiovascular disease risk.
Figure 5
Overview of the experimental procedure and the use of datasets.
We used a threefold within-sample (DyNAMiC) cross-validation where we trained our model on 120 subjects (8:2; 80% training:20% validation during training) and tested it in a separate sample of 60 subjects. The winning within-sample model was used for between-sample (COBRA) external validation.
Figure 6
DenselyAttention network architecture for functional connectivity analysis across cognitive states.
(a) Example of a functional connectivity map across three different cognitive states. SEN hand: SENsory hand; SEN Mouth: SENsory Mouth; CON: cingulo-operculum control network; Aud: auditory; DMN: default mode network; Memory Ret: memory retrieval network; Visual: Visual; FPN: fronto-parietal network; Salience: Salience control network; Subcortical (upper row): subcortical network included in original power parcellation; VAN: ventral attention network; DAN: dorsal attention network; Cerebellar: cerebellar network; Subcortical (lower row): additional subcortical regions, including hippocampus and caudate, added to the original power parcellation; Uncertain: regions with less known network assignment. (b) DenselyAttention architecture. enhanced residual block (ERB) and high-frequency attention block (HFAB) into the transition block. Note that each ‘D.L.’ layer in the table corresponds to the sequence BatchNormalization-ReLU-Conv3×3.
Tables
Table 1
Correlation results for episodic memory (EM) score predictions.
| Model trained on | rest | movie | n-back | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Model tested on | rest | movie | n-back | rest | movie | n-back | rest | movie | n-back |
| Correlation (r) | 0.50 | 0.44 | 0.38 | 0.28 | 0.49 | 0.09 | 0.05 | 0.08 | 0.17 |
| Correlation (r2) | 0.13 | 0.05 | 0.04 | 0.03 | 0.12 | 0.00 | –0.01 | 0.00 | 0.02 |
| Significance | <0.0001 | 0.0004 | 0.003 | 0.02 | <0.0001 | 0.49 | 0.76 | 0.84 | 0.42 |
| MSE MAE | 4.35 1.59 | 81.93 3.87 | 40.74 4.92 | 129.44 11.88 | 10.83 2.57 | 156.42 8.75 | 133.87 9.33 | 127.57 8.28 | 20.16 3.42 |
| COBRA | |||||||||
| Correlation (r) | 0.24 | ||||||||
| Correlation (r2) | 0.05 | ||||||||
| Significance | <0.0001 | ||||||||
| MSE MAE | 37.62 8.28 | ||||||||
Table 2
Correlation results for working memory (WM) score predictions.
| Model trained on | rest | movie | n-back | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Model tested on | rest | movie | n-back | rest | movie | n-back | rest | movie | n-back |
| Correlation (r) | 0.02 | 0.08 | 0.12 | 0.42 | 0.57 | 0.46 | 0.20 | 0.46 | 0.47 |
| Correlation (r2) | –0.09 | –0.05 | –0.03 | 0.11 | 0.17 | 0.09 | 0.02 | 0.05 | 0.14 |
| Significance MSE MAE | 0.88 85.67 7.47 | 0.72 214.81 9.74 | 0.15 169.02 9.37 | <0.0001 70.57 6.67 | <0.0001 49.19 5.70 | 0.004 98.36 8.10 | 0.12 127.72 8.89 | 0.0002 116.15 8.71 | <0.0001 61.23 6.31 |
| COBRA | |||||||||
| Correlation (r) | 0.47 | ||||||||
| Correlation (r2) | 0.10 | ||||||||
| Significance MSE MAE | <0.0001 94.87 7.65 | ||||||||
Additional files
-
MDAR checklist
- https://cdn.elifesciences.org/articles/104053/elife-104053-mdarchecklist1-v1.pdf
-
Supplementary file 1
Replication of correlation results for EM and WM predictions using Schaefer-300 parcellation.
- https://cdn.elifesciences.org/articles/104053/elife-104053-supp1-v1.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Brain cognition gaps reveal associations with dopamine and factors related to brain health through artificial intelligence prediction of functional connectome
eLife 14:RP104053.
https://doi.org/10.7554/eLife.104053.4
