Balint Kincses
Research Subjects
Balint Kincses has a total of 86 co-authors. Here are the top co-authors by number of shared publications:
- Tamas Spisak
University Hospital Essen
13 shared publications - Ulrike Bingel
Department of Neurology, Center for Translational Neuro- and Behavioral Sciences (C-TNBS), University Medicine Essen, University Duisburg-Essen
13 shared publications - Raviteja Kotikalapudi
Center for Translational Neuro- and Behavioral Sciences (C-TNBS), University Medicine Essen
8 shared publications - Giuseppe Gallitto
Center for Translational Neuro- and Behavioral Sciences (C-TNBS), University Medicine Essen
7 shared publications - Robert Englert
Center for Translational Neuro- and Behavioral Sciences (C-TNBS), University Medicine Essen
6 shared publications - Jialin Li
Center for Translational Neuro- and Behavioral Sciences (C-TNBS), University Medicine Essen
4 shared publications
Latest Publications
- Open Access
Comparison of MRI sequences for optic nerve lesion detection in the follow-up of multiple sclerosis
Background: Optic nerve involvement is common in multiple sclerosis (MS) and is now recognized as a key site for dissemination in space under the most recent revision of McDonald’s criteria. Reliable detection of optic nerve lesions is essential for diagnosis and monitoring, yet the optimal MRI sequence remains uncertain. Objective: To compare the diagnostic performance of three MRI sequences—short tau inversion recovery (STIR), fat-suppressed FLAIR (fs-FLAIR), and double inversion recovery (DIR)—in detecting optic nerve lesions in MS patients. Methods: Fifty-nine MS patients underwent MRI with STIR, fs-FLAIR, and DIR sequences and visual evoked potential (VEP) testing. Lesion detection was assessed independently for each sequence, and results were compared to structural and functional standards. Results: No significant differences were found in lesion detection across the three sequences. All sequences showed similar sensitivity to structural and functional changes. The incremental benefit of adding orbita specific sequence to a whole-brain sequence was limited in the follow-up of MS. Conclusion: In patients with established MS, whole-brain sequences (fs-FLAIR, DIR) perform comparably to dedicated orbital sequences (STIR) in detecting optic nerve lesions. This supports the feasibility of MRI protocols by omitting additional orbital sequences in routine follow-up, thereby reducing scan time and patient burden without compromising diagnostic sensitivity.
Open AccessComparison of MRI sequences for optic nerve lesion detection in the follow-up of multiple sclerosis
Background: Optic nerve involvement is common in multiple sclerosis (MS) and is now recognized as a key site for dissemination in space under the most recent revision of McDonald’s criteria. Reliable detection of optic nerve lesions is essential for diagnosis and monitoring, yet the optimal MRI sequence remains uncertain. Objective: To compare the diagnostic performance of three MRI sequences—short tau inversion recovery (STIR), fat-suppressed FLAIR (fs-FLAIR), and double inversion recovery (DIR)—in detecting optic nerve lesions in MS patients. Methods: Fifty-nine MS patients underwent MRI with STIR, fs-FLAIR, and DIR sequences and visual evoked potential (VEP) testing. Lesion detection was assessed independently for each sequence, and results were compared to structural and functional standards. Results: No significant differences were found in lesion detection across the three sequences. All sequences showed similar sensitivity to structural and functional changes. The incremental benefit of adding orbita specific sequence to a whole-brain sequence was limited in the follow-up of MS. Conclusion: In patients with established MS, whole-brain sequences (fs-FLAIR, DIR) perform comparably to dedicated orbital sequences (STIR) in detecting optic nerve lesions. This supports the feasibility of MRI protocols by omitting additional orbital sequences in routine follow-up, thereby reducing scan time and patient burden without compromising diagnostic sensitivity.
- Open Access
Beyond pain: a brain-based biomarker predicts individual pain relief
Effective pain relief is a central goal of medical care, yet objective biomarkers of pain relief are lacking. Using task-based functional MRI and a capsaicin-induced tonic heat pain model, we experimentally elicited both pain exacerbation and pain relief within the same individuals. Existing brain-based signatures, including the Neurologic Pain Signature (NPS), reliably detected pain increases, but failed to capture pain relief. We therefore developed the PAin and RElief Signature (PARES), a multivariate brain signature trained to predict bidirectional changes in pain perception in n = 61 healthy controls. PARES robustly predicted both pain increases and relief and generalized to an independent cohort of people with chronic back pain (n = 58), who underwent the same experimental procedures. Together these findings establish a neural signature of pain relief and provide a potential biomarker for treatment stratification and analgesic development.
Open AccessBeyond pain: a brain-based biomarker predicts individual pain relief
Effective pain relief is a central goal of medical care, yet objective biomarkers of pain relief are lacking. Using task-based functional MRI and a capsaicin-induced tonic heat pain model, we experimentally elicited both pain exacerbation and pain relief within the same individuals. Existing brain-based signatures, including the Neurologic Pain Signature (NPS), reliably detected pain increases, but failed to capture pain relief. We therefore developed the PAin and RElief Signature (PARES), a multivariate brain signature trained to predict bidirectional changes in pain perception in n = 61 healthy controls. PARES robustly predicted both pain increases and relief and generalized to an independent cohort of people with chronic back pain (n = 58), who underwent the same experimental procedures. Together these findings establish a neural signature of pain relief and provide a potential biomarker for treatment stratification and analgesic development.
- Open Access
Reinforcement Learning via Brain Feedback for real-time fMRI-based adaptive stimulus generation
Traditional fMRI studies rely on predefined task paradigms, where fixed stimulus designs limit the flexibility with which brain-stimulus relationships can be explored. Here, we introduce Reinforcement Learning via Brain Feedback (RLBF), a framework and open-source software package for adaptive stimulus optimization using real-time fMRI. RLBF reverses the conventional direction of inference by using neural responses to guide the exploration of stimulus spaces through reinforcement learning, enabling optimization of predefined brain targets such as regional activity or multivariate neural signatures. The accompanying Python-based software provides a modular framework integrating real-time fMRI data processing, reinforcement learning agents, adaptive stimulus generation, simulation-based testing, and experiment monitoring. Its flexible architecture allows researchers to customize preprocessing pipelines, reward functions, stimulus spaces, and RL strategies for diverse closed-loop neuroimaging applications. We validate the framework in a proof-of-concept study (N=10), demonstrating real-time optimization of a simple visual stimulus space by adapting checkerboard contrast and frequency to maximize primary visual cortex (V1) responses within a single 10-minute fMRI session. RLBF provides an extensible foundation for brain-guided stimulus optimization and enables new approaches for investigating neural specificity, individualized brain–stimulus relationships, and adaptive experimental design.
Open AccessReinforcement Learning via Brain Feedback for real-time fMRI-based adaptive stimulus generation
Traditional fMRI studies rely on predefined task paradigms, where fixed stimulus designs limit the flexibility with which brain-stimulus relationships can be explored. Here, we introduce Reinforcement Learning via Brain Feedback (RLBF), a framework and open-source software package for adaptive stimulus optimization using real-time fMRI. RLBF reverses the conventional direction of inference by using neural responses to guide the exploration of stimulus spaces through reinforcement learning, enabling optimization of predefined brain targets such as regional activity or multivariate neural signatures. The accompanying Python-based software provides a modular framework integrating real-time fMRI data processing, reinforcement learning agents, adaptive stimulus generation, simulation-based testing, and experiment monitoring. Its flexible architecture allows researchers to customize preprocessing pipelines, reward functions, stimulus spaces, and RL strategies for diverse closed-loop neuroimaging applications. We validate the framework in a proof-of-concept study (N=10), demonstrating real-time optimization of a simple visual stimulus space by adapting checkerboard contrast and frequency to maximize primary visual cortex (V1) responses within a single 10-minute fMRI session. RLBF provides an extensible foundation for brain-guided stimulus optimization and enables new approaches for investigating neural specificity, individualized brain–stimulus relationships, and adaptive experimental design.
- Open Access
- 17 July 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1) , 6538
Meta-analytic evidence for distinct neural correlates of conditioned versus verbally induced placebo analgesia
Placebo analgesia demonstrates that belief and expectation can significantly alter pain, even without active treatment. Placebo analgesia can be induced through verbal suggestion, classical conditioning, or their combination, though the role of conditioned neural responses above and beyond effects of verbal instructions remains unclear. We conduct a systematic meta-analysis of individual participant data from 16 within-participant placebo neuroimaging studies ( n = 409), employing univariate and multivariate analyses to identify shared and distinct mechanisms of placebo analgesia induced by suggestions alone versus suggestions combined with conditioning. Both techniques increase activity during pain in the dorsolateral prefrontal and inferior parietal cortices and decrease activation in the insula, putamen, and primary sensory areas. Adding conditioning enhances engagement of regions associated with context representation and pain modulation (e.g., dorsolateral/dorsomedial prefrontal cortices) and decreases in nociceptive regions (e.g., primary sensory and insular areas). Conditioning also strengthens the negative association between analgesia and nociceptive activity, as quantified by the Neurologic Pain Signature. Combining conditioning with instructions yields greater placebo analgesia, mediated by increased ventromedial prefrontal and dorsal caudate activity, alongside decreased sensory-nociceptive and cerebellar activity. These findings suggest the two strategies rely on partially distinct mechanisms in the brain.
- 17 July 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1) , 6538
Open AccessMeta-analytic evidence for distinct neural correlates of conditioned versus verbally induced placebo analgesia
Placebo analgesia demonstrates that belief and expectation can significantly alter pain, even without active treatment. Placebo analgesia can be induced through verbal suggestion, classical conditioning, or their combination, though the role of conditioned neural responses above and beyond effects of verbal instructions remains unclear. We conduct a systematic meta-analysis of individual participant data from 16 within-participant placebo neuroimaging studies ( n = 409), employing univariate and multivariate analyses to identify shared and distinct mechanisms of placebo analgesia induced by suggestions alone versus suggestions combined with conditioning. Both techniques increase activity during pain in the dorsolateral prefrontal and inferior parietal cortices and decrease activation in the insula, putamen, and primary sensory areas. Adding conditioning enhances engagement of regions associated with context representation and pain modulation (e.g., dorsolateral/dorsomedial prefrontal cortices) and decreases in nociceptive regions (e.g., primary sensory and insular areas). Conditioning also strengthens the negative association between analgesia and nociceptive activity, as quantified by the Neurologic Pain Signature. Combining conditioning with instructions yields greater placebo analgesia, mediated by increased ventromedial prefrontal and dorsal caudate activity, alongside decreased sensory-nociceptive and cerebellar activity. These findings suggest the two strategies rely on partially distinct mechanisms in the brain.
- Open Access
Layer-specific cortical processing dissociates sensory and cognitive influences on pain
Pain arises from the integration of nociceptive input with cognitive and affective processes, yet how these signals are organized within cortical circuits remains unclear. Here, using submillimeter-resolution 7T fMRI, we tested whether bottom-up (BU) and top-down (TD) influences on pain are segregated across cortical layers in human sensory-discriminative regions. Participants underwent a factorial manipulation of nociceptive input and cognitive modulation, enabling dissociation of BU and TD processes. BU processing was strongest in middle cortical layers. In contrast, TD modulation preferentially engaged superficial layers, consistent with cortico-cortical feedback mechanisms. Critically, individual differences in this laminar segregation predicted the magnitude of distraction-induced analgesia, linking circuit-level organization to behavior. These findings provide evidence that cognitive modulation of pain is implemented through layer-specific cortical computations and extend canonical microcircuit models to human pain processing. More generally, they establish laminar fMRI as a powerful approach for linking cortical circuit architecture to subjective experience.
Open AccessLayer-specific cortical processing dissociates sensory and cognitive influences on pain
Pain arises from the integration of nociceptive input with cognitive and affective processes, yet how these signals are organized within cortical circuits remains unclear. Here, using submillimeter-resolution 7T fMRI, we tested whether bottom-up (BU) and top-down (TD) influences on pain are segregated across cortical layers in human sensory-discriminative regions. Participants underwent a factorial manipulation of nociceptive input and cognitive modulation, enabling dissociation of BU and TD processes. BU processing was strongest in middle cortical layers. In contrast, TD modulation preferentially engaged superficial layers, consistent with cortico-cortical feedback mechanisms. Critically, individual differences in this laminar segregation predicted the magnitude of distraction-induced analgesia, linking circuit-level organization to behavior. These findings provide evidence that cognitive modulation of pain is implemented through layer-specific cortical computations and extend canonical microcircuit models to human pain processing. More generally, they establish laminar fMRI as a powerful approach for linking cortical circuit architecture to subjective experience.
- Open Access
- 18 March 2026
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 13
Functional connectivity-based attractor dynamics of the human brain in rest, task, and disease
Functional brain connectivity has been instrumental in uncovering the large-scale organization of the brain and its relation to various behavioral and clinical phenotypes. Understanding how this functional architecture relates to the brain’s dynamic activity repertoire is an essential next step towards interpretable generative models of brain function. We propose functional connectivity-based Attractor Neural Networks (fcANNs), a theoretically inspired model of macro-scale brain dynamics, simulating recurrent activity flow among brain regions based on first principles of self-organization. In the fcANN framework, brain dynamics are understood in relation to attractor states; neurobiologically meaningful activity configurations that minimize the free energy of the system. We provide the first evidence that large-scale brain attractors - as reconstructed by fcANNs - exhibit an approximately orthogonal organization, which is a signature of the self-orthogonalization mechanism of the underlying theoretical framework of free-energy-minimizing attractor networks. Analyses of seven distinct human neuroimaging datasets demonstrate that fcANNs can accurately reconstruct and predict brain dynamics under a wide range of conditions, including resting and task states, and brain disorders. By establishing a formal link between connectivity and activity, fcANNs offer a simple and interpretable computational alternative to conventional descriptive analyses.
- 18 March 2026
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 13
Open AccessFunctional connectivity-based attractor dynamics of the human brain in rest, task, and disease
Functional brain connectivity has been instrumental in uncovering the large-scale organization of the brain and its relation to various behavioral and clinical phenotypes. Understanding how this functional architecture relates to the brain’s dynamic activity repertoire is an essential next step towards interpretable generative models of brain function. We propose functional connectivity-based Attractor Neural Networks (fcANNs), a theoretically inspired model of macro-scale brain dynamics, simulating recurrent activity flow among brain regions based on first principles of self-organization. In the fcANN framework, brain dynamics are understood in relation to attractor states; neurobiologically meaningful activity configurations that minimize the free energy of the system. We provide the first evidence that large-scale brain attractors - as reconstructed by fcANNs - exhibit an approximately orthogonal organization, which is a signature of the self-orthogonalization mechanism of the underlying theoretical framework of free-energy-minimizing attractor networks. Analyses of seven distinct human neuroimaging datasets demonstrate that fcANNs can accurately reconstruct and predict brain dynamics under a wide range of conditions, including resting and task states, and brain disorders. By establishing a formal link between connectivity and activity, fcANNs offer a simple and interpretable computational alternative to conventional descriptive analyses.
- Open Access
- 21 November 2025
- Published by Springer Nature in Communications Biology
- Vol. 8 (1) , 1628
Author Correction: On the replicability of diffusion weighted MRI-based brain-behavior models
- 21 November 2025
- Published by Springer Nature in Communications Biology
- Vol. 8 (1) , 1628
Open AccessAuthor Correction: On the replicability of diffusion weighted MRI-based brain-behavior models
- Open Access
- 30 October 2025
- Published by Springer Nature in Communications Biology
- Vol. 8 (1) , 1-8
On the replicability of diffusion weighted MRI-based brain-behavior models
Replicability of anatomical and functional MRI-based inter-individual BWAS has been extensively discussed recently. This study reports a comprehensive evaluation of BWAS replicability based on structural connectomes (streamlines, FA, MD, RD, AD). Overall, 36%(21/58) of brain-phenotype associations were replicable in the HCP dataset with atleast one of the DWI metrics, and a discovery sample size n ≤ 425 (total sample size = discovery + replication samples). Temporally stable, trait-like phenotypes were found to be more replicable (50%, 16/32), than state-like measures (19%, 5/26). Streamline-based connectomes (SC) provided the highest replicability across all metrics (29% and 42% of phenotypes in the HCP and AOMIC datasets, respectively). In line with theoretical expectations, replicability was found to be directly related to effect size. Phenotypes that needed n > 400 discovery samples to replicate displayed very low effect sizes 425 samples will necessarily display limited practical relevance due to their small predictive performance. Large sample sizes remain crucial for explainability and for assessing fairness and generalizability to new populations.
- 30 October 2025
- Published by Springer Nature in Communications Biology
- Vol. 8 (1) , 1-8
Open AccessOn the replicability of diffusion weighted MRI-based brain-behavior models
Replicability of anatomical and functional MRI-based inter-individual BWAS has been extensively discussed recently. This study reports a comprehensive evaluation of BWAS replicability based on structural connectomes (streamlines, FA, MD, RD, AD). Overall, 36%(21/58) of brain-phenotype associations were replicable in the HCP dataset with atleast one of the DWI metrics, and a discovery sample size n ≤ 425 (total sample size = discovery + replication samples). Temporally stable, trait-like phenotypes were found to be more replicable (50%, 16/32), than state-like measures (19%, 5/26). Streamline-based connectomes (SC) provided the highest replicability across all metrics (29% and 42% of phenotypes in the HCP and AOMIC datasets, respectively). In line with theoretical expectations, replicability was found to be directly related to effect size. Phenotypes that needed n > 400 discovery samples to replicate displayed very low effect sizes 425 samples will necessarily display limited practical relevance due to their small predictive performance. Large sample sizes remain crucial for explainability and for assessing fairness and generalizability to new populations.
- Open Access
Meta-analytic evidence for distinct neural correlates of conditioned vs. verbally induced placebo analgesia
Placebo analgesia demonstrates that belief and expectation can significantly alter pain, even without active treatment. Placebo analgesia can be induced through verbal suggestion, classical conditioning, or their combination, though the role of conditioned neural responses above and beyond effects of verbal instructions remains unclear. We conducted a systematic meta-analysis of individual participant data from 16 within-participant placebo neuroimaging studies ( n = 409), employing univariate and multivariate analyses to identify shared and distinct mechanisms of placebo analgesia induced by suggestions alone versus suggestions combined with conditioning. Both techniques increased activity during pain in the dorsolateral prefrontal and inferior parietal cortices and decreased activation in the insula, putamen, and primary sensory areas. Adding conditioning enhanced engagement of regions associated with context representation and pain modulation (e.g., dorsolateral/dorsomedial prefrontal cortices) and decreases in nociceptive regions (e.g., primary sensory and insular areas). Conditioning also strengthened the association between analgesia and nociceptive activity, as reflected in the Neurologic Pain Signature. Combining conditioning with instructions yielded greater analgesia, mediated by increased ventromedial prefrontal and dorsal caudate activity, alongside decreased sensory-nociceptive and cerebellar activity. These findings suggest the two strategies rely on partially distinct mechanisms, which could be combined to optimize placebo analgesia’s clinical application.
Open AccessMeta-analytic evidence for distinct neural correlates of conditioned vs. verbally induced placebo analgesia
Placebo analgesia demonstrates that belief and expectation can significantly alter pain, even without active treatment. Placebo analgesia can be induced through verbal suggestion, classical conditioning, or their combination, though the role of conditioned neural responses above and beyond effects of verbal instructions remains unclear. We conducted a systematic meta-analysis of individual participant data from 16 within-participant placebo neuroimaging studies ( n = 409), employing univariate and multivariate analyses to identify shared and distinct mechanisms of placebo analgesia induced by suggestions alone versus suggestions combined with conditioning. Both techniques increased activity during pain in the dorsolateral prefrontal and inferior parietal cortices and decreased activation in the insula, putamen, and primary sensory areas. Adding conditioning enhanced engagement of regions associated with context representation and pain modulation (e.g., dorsolateral/dorsomedial prefrontal cortices) and decreases in nociceptive regions (e.g., primary sensory and insular areas). Conditioning also strengthened the association between analgesia and nociceptive activity, as reflected in the Neurologic Pain Signature. Combining conditioning with instructions yielded greater analgesia, mediated by increased ventromedial prefrontal and dorsal caudate activity, alongside decreased sensory-nociceptive and cerebellar activity. These findings suggest the two strategies rely on partially distinct mechanisms, which could be combined to optimize placebo analgesia’s clinical application.
- Top Cited PapersOpen Access
- 6 January 2025
- Published by Oxford University Press (OUP) in GigaScience
- Vol. 14
External validation of machine learning models—registered models and adaptive sample splitting
Background: Multivariate predictive models play a crucial role in enhancing our understanding of complex biological systems and in developing innovative, replicable tools for translational medical research. However, the complexity of machine learning methods and extensive data preprocessing and feature engineering pipelines can lead to overfitting and poor generalizability. An unbiased evaluation of predictive models necessitates external validation, which involves testing the finalized model on independent data. Despite its importance, external validation is often neglected in practice due to the associated costs. Results: Here we propose that, for maximal credibility, model discovery and external validation should be separated by the public disclosure (e.g., preregistration) of feature processing steps and model weights. Furthermore, we introduce a novel approach to optimize the trade-off between efforts spent on model discovery and external validation in such studies. We show on data involving more than 3,000 participants from four different datasets that, for any “sample size budget,” the proposed adaptive splitting approach can successfully identify the optimal time to stop model discovery so that predictive performance is maximized without risking a low-powered, and thus inconclusive, external validation. Conclusion: The proposed design and splitting approach (implemented in the Python package “AdaptiveSplit”) may contribute to addressing issues of replicability, effect size inflation, and generalizability in predictive modeling studies.
- 6 January 2025
- Published by Oxford University Press (OUP) in GigaScience
- Vol. 14
Top Cited PapersOpen AccessExternal validation of machine learning models—registered models and adaptive sample splitting
Background: Multivariate predictive models play a crucial role in enhancing our understanding of complex biological systems and in developing innovative, replicable tools for translational medical research. However, the complexity of machine learning methods and extensive data preprocessing and feature engineering pipelines can lead to overfitting and poor generalizability. An unbiased evaluation of predictive models necessitates external validation, which involves testing the finalized model on independent data. Despite its importance, external validation is often neglected in practice due to the associated costs. Results: Here we propose that, for maximal credibility, model discovery and external validation should be separated by the public disclosure (e.g., preregistration) of feature processing steps and model weights. Furthermore, we introduce a novel approach to optimize the trade-off between efforts spent on model discovery and external validation in such studies. We show on data involving more than 3,000 participants from four different datasets that, for any “sample size budget,” the proposed adaptive splitting approach can successfully identify the optimal time to stop model discovery so that predictive performance is maximized without risking a low-powered, and thus inconclusive, external validation. Conclusion: The proposed design and splitting approach (implemented in the Python package “AdaptiveSplit”) may contribute to addressing issues of replicability, effect size inflation, and generalizability in predictive modeling studies.