Generalizing prediction of task-evoked brain activity across datasets and populations

Tik, Niv; Gal, Shachar; Madar, Asaf; Ben-David, T.; Bernstein‐Eliav, Michal; Tavor, Ido

doi:10.1016/j.neuroimage.2023.120213

Cited by 6 publications

(8 citation statements)

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Feature extraction for resting-state functional modes Previous studies have used resting-state functional modes for predicting task activation maps (Gal, Coldham, et al, 2022;Tavor et al, 2016;Tik et al, 2021Tik et al, , 2023Zheng et al, 2022). Functional modes refer to consistent spatial patterns of brain activity observed across different individuals.…”

Section: Methodsmentioning

confidence: 99%

“…Such fingerprints have been shown to predict task activation across various cognitive tasks (Cohen et al, 2020;Cole et al, 2016;Tavor et al, 2016;Tripathi & Somers, 2023;Zheng et al, 2022), even in patients with neurological disorders (Niu et al, 2021;Parker Jones et al, 2017;Tik et al, 2021). The consistent correlation between resting-state functional connectivity and task activation was validated across diverse sites, MRI vendors, and age groups in multiple cognitive tasks (Tik et al, 2023). In addition, task activation maps derived by resting-state functional connectivity achieved more accurate predictions of cognitive functioning than resting-state functional connectivity alone Furthermore, researchers can investigate underlying functional connections of task-related cognitive processes by identifying resting-state functional connectivity predictive of task-induced brain activity (Izakson et al, 2023;Tik et al, 2021).…”

Section: Introductionmentioning

confidence: 90%

“…The advent of deep learning marks a significant evolution in connectivity fingerprinting from fMRI data, enabling enhanced predictive accuracy and reliability over traditional methods. BrainSurfCNN, a surfacebased deep convolutional neural network pre-trained on extensive datasets and fine-tuned for specific applications, has shown marked capability in predicting task activation maps from resting-state gray-ordinate Coldham, et al, 2022;Tavor et al, 2016;Tik et al, 2021Tik et al, , 2023. Specifically, we employed 100 cortical parcels defined by Schaefer et al, 2018, which are assigned to one of the seven brain networks and widely used in the previous research (Cohen et al, 2020;Gal, Coldham, et al, 2022;Tavor et al, 2016), and Harvard-Oxford cortical and subcortical structural atlases for evaluating the model's performance in predicting subcortical brain activity (Desikan et al, 2006;Frazier et al, 2005;Goldstein et al, 2007;Makris et al, 2006).…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, we used three metrics to compare how well each model captures individual differences in task activation maps: the diagonality index, the diagonal percentile mean, and the effect size (D) of the Kolmogorov-Smirnov test. The diagonality index, a commonly used metric in previous studies (Tik et al, 2021(Tik et al, , 2023Zheng et al, 2022), is a mean of off-diagonal correlations subtracted from the mean of diagonal correlations. To evaluate whether a task activation map generated by the model for a particular subject specifically predicts the actual activation map for that subject compared to task activation maps generated for other subjects, we developed a new metric, termed the diagonal percentile mean.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Predicting task-related brain activity from resting-state brain dynamics with fMRI Transformer

Kwon,

Seo,

Wang

et al. 2024

Preprint

View full text Add to dashboard Cite

Accurate prediction of the brain's task reactivity from resting-state functional magnetic resonance imaging (fMRI) data remains a significant challenge in neuroscience. Traditional statistical approaches often fail to capture the complex, nonlinear spatiotemporal patterns of brain function. This study introduces SwiFUN (Swin fMRI UNet Transformer), a novel deep-learning framework designed to predict 3D task activation maps directly from resting-state fMRI scans. SwiFUN leverages advanced techniques such as shifted window-based self-attention, which helps to understand complex patterns by focusing on varying parts of the data sequentially, and a contrastive learning strategy to better capture individual differences among subjects. When applied to predicting emotion-related task activation in adults (UK Biobank, n=7,038) and children (ABCD, n=4,944), SwiFUN consistently achieved higher overall prediction accuracy than existing methods across all contrasts; it demonstrated an improvement of up to 27% for the FACES-PLACES contrast in ABCD data. The resulting task activation maps revealed individual differences across cortical and subcortical regions associated with sex, age, and depressive symptoms. This scalable, transformer-based approach potentially reduces the need for task-based fMRI in clinical settings, marking a promising direction for future neuroscience and clinical research that enhances our ability to understand and predict brain function.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 90%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Predicting task-related brain activity from resting-state brain dynamics with fMRI Transformer

Kwon,

Seo,

Wang

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Tavor et al, 2016;Tik et al, 2023Tik et al, , 2021 and the recent suggestion that functional network architecture may relate brain activity and behaviour (Bernstein-Eliav and Tavor, 2022; Bijsterbosch et al, 2020), a network-organization approach may facilitate the study of neuroplasticity and its behavioural manifestation as well.…”

mentioning

confidence: 99%

Continuous structural neuroplasticity during motor learning - a diffusion MRI study

Friedman,

Malovani,

Perets

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

Learning is a complex, continuously evolving process which induces widespread structural and functional brain alterations. In recent years, diffusion magnetic resonance imaging (dMRI) has been used to detect microstructural modifications after short learning periods. While previous studies primarily explored changesfollowinglearning (i.e., by comparing images acquired before and after learning), the continuous temporal dynamics of microstructural alterationsduringthe encoding phase of learning itself are yet to be examined. Here, we introduce a novel approach for continuous acquisition of dMRI images which allow tracking microstructural changes throughout learning and demonstrate the utility of this approach on a motor sequence learning (finger tapping) task (n=58). Voxel-wise analysis revealed a decrease in mean diffusivity (MD) in task-related brain regions, including the parahippocampal gyrus, hippocampus, inferior temporal gyrus, and cerebellum. Further analysis of the temporal patterns of decrease revealed a rapid MD reduction in the right temporal gyrus after 11 minutes of learning, with additional decrease in the right parahippocampal gyrus and left cerebellum after 22 minutes. We computed "neuroplasticity networks" of brain areas showing similar change patterns and detected similarities between these networks and canonical functional connectivity networks. Our findings offer novel insights on the spatio-temporal dynamics of neuroplasticity and advance our understanding of motor learning by demonstrating continuous microstructural modifications from the very first minutes of training.

show abstract

TS-AI: A deep learning pipeline for multimodal subject-specific parcellation with task contrasts synthesis

Li,

Lu,

et al. 2024

Medical Image Analysis

View full text Add to dashboard Cite

Generalizing prediction of task-evoked brain activity across datasets and populations

Cited by 6 publications

References 60 publications

Predicting task-related brain activity from resting-state brain dynamics with fMRI Transformer

Predicting task-related brain activity from resting-state brain dynamics with fMRI Transformer

Continuous structural neuroplasticity during motor learning - a diffusion MRI study

TS-AI: A deep learning pipeline for multimodal subject-specific parcellation with task contrasts synthesis

Contact Info

Product

Resources

About