MBMarlene Bönstrup

Marlene Bönstrup

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Marlene Bönstrup has a total of 89 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    The structural reserve of brain networks influences outcomes after a stroke

    by Lukas Frontzkowski, Tim J Hunze, Winifried Backhaus, Marlene Bönstrup, Christian Gerloff, Bastian Cheng, Götz Thomalla, Benedikt M Frey, Paweł P Wróbel, Hanna Braaß, et al.

    Structural brain reserve capacity has recently gained an increasing interest in stroke recovery research. Focal and global measures of brain reserve have been linked with recovery trajectories. Whether the reserve localized within large-scale brain networks might also carry information to better understand outcome variability after stroke is an open question. This work analysed 31 patients with severe, first-ever unilateral, supratentorial stroke. Patients underwent MRI brain imaging and clinical testing within the first 2 weeks after the event and a longitudinal clinical follow-up after 3–6 months. Individual tractography in the contralesional hemisphere was performed to reconstruct structural connectomes to approximate the state of the ipsilesional brain networks before the stroke. Graph theory was applied to describe network integration and segregation topologies. Linear and ordinal logistic regression analyses were conducted to associate network topologies at baseline with neurological symptom burden, global and activity-related disability and motor impairment at follow-up. The main finding was that less segregated and more integrated networks, characterized by lower network modularity and higher efficiency, were linked with a more favourable outcome on follow-up. Modularity exerted a remarkably consistent influence across various outcome measures. This association was independent of the initial deficit, lesion volume or age. This study sheds novel light on brain reserve, localizing within the topology of pre-stroke structural brain networks, as a critical determinant of recovery after stroke.


  • Open Access

    Group-derived and individual disconnection in stroke: recovery prediction and deep graph learning

    by Patrik Bey, Kiret Dhindsa, Torsten Rackoll, Jan Feldheim, Marlene Bönstrup, Götz Thomalla, Robert Schulz, Bastian Cheng, Christian Gerloff, Matthias Endres, et al.

    Recent advances in the treatment of acute ischemic stroke contribute to improved patient outcomes, yet the mechanisms driving long-term disease trajectory are not well-understood. Current trends in the literature emphasize the distributed disruptive impact of stroke lesions on brain network organization. While most studies use population-derived data to investigate lesion interference on healthy tissue, the potential for individualized treatment strategies remains underexplored due to a lack of availability and effective utilization of the necessary clinical imaging data. To validate the potential for individualized patient evaluation, we explored and compared the differential information in network models based on normative and individual data. We further present our novel deep learning approach providing usable and accurate estimates of individual stroke impact utilizing minimal imaging data, thus bridging the data gap hindering individualized treatment planning. We created normative and individual disconnectomes for each of 78 patients (mean age 65.1 years, 32 females) from two independent cohort studies. MRI data and Barthel Index, as a measure of activities of daily living, were collected in the acute and early sub-acute phase after stroke (baseline) and at three months post stroke incident. Disconnectomes were subsequently described using 12 network metrics, including clustering coefficient and transitivity. Metrics were first compared between disconnectomes and further utilized as features in a classifier to predict a patients’ disease trajectory, as defined by three months Barthel Index. We then developed a deep learning architecture based on graph convolution and trained it to predict properties of the individual disconnectomes from the normative disconnectomes. Both disconnectomes showed statistically significant differences in topology and predictive power. Normative disconnectomes included a statistically significant larger number of connections (N=604 for normative versus N=210 for individual) and agreement between network properties ranged from r 2 =0.01 for clustering coefficient to r 2 =0.8 for assortativity, highlighting the impact of disconnectome choice on subsequent analysis. To predict patient deficit severity, individual data achieved an AUC score of 0.94 compared to an AUC score of 0.85 for normative based features. Our deep learning estimates showed high correlation with individual features (mean r 2 =0.94) and a comparable performance with an AUC score of 0.93. We were able to show how normative data-based analysis of stroke disconnections provides limited information regarding patient recovery. In contrast, individual data provided higher prognostic precision. We presented a novel approach to curb the need for individual data while retaining most of the differential information encoding individual patient disease trajectory.


  • Open Access

    Sequence action representations contextualize during early skill learning

    by Debadatta Dash, Fumiaki Iwane, William Hayward, Roberto Salamanca-Giron, Marlene Bonstrup, Ethan R Buch, Leonardo G Cohen

    Activities of daily living rely on our ability to acquire new motor skills composed of precise action sequences. Here, we asked if the millisecond-level neural representation of an action performed at different contextual sequence locations within a skill differentiates or remains stable during early motor learning. We first optimized machine learning decoders predictive of sequence-embedded finger movements from magnetoencephalographic (MEG) activity. Using this approach, we found that the neural representation of the same action performed in different contextual sequence locations, progressively differentiated—primarily during rest intervals of early learning (offline)—correlating with skill gains. In contrast, representational differentiation during practice (online) did not reflect learning. The regions contributing to this representational differentiation evolved with learning, shifting from the contralateral pre- and post-central cortex during early learning (trials 1–11) to increased involvement of the superior and middle frontal cortex once skill performance plateaued (trials 12–36). Thus, the neural substrates supporting finger movements and their representational differentiation during early skill learning differ from those supporting stable performance during the subsequent skill plateau period. Representational contextualization extended to Day 2, exhibiting specificity for the practiced skill sequence. Altogether, our findings indicate that sequence action representations contextually differentiate during early skill learning, an issue relevant to brain-computer interface applications in neurorehabilitation.

  • Neurotransmitter-informed connectivity maps and their application for outcome inference after stroke

    by Philipp J Koch, Benedikt M Frey, Winifried Backhaus, Nora Petersen, Gabriel Girard, Pawel P Wróbel, Hanna Braaß, Marlene Bönstrup, Lisa Kunkel Genannt Bode, Bastian Cheng, et al.

    Neuroscience has evolved by framing numerous neuropsychiatric conditions as network diseases. Alterations within neurotransmitter (NT) systems are central to the development of these diseases. Recently, normative data on whole-brain NT fingerprints derived from PET tracer data have become accessible; limited data related this information to sequelae after stroke. This work aimed to explore: (i) the integration of NT data into whole-brain structural connectivity analyses; and (ii) its potential contribution to understanding outcome variability following stroke. Normative maps of NT receptor and transporter densities were integrated with a normative structural connectome to generate NT-specific connectivity maps for serotonin, dopamine, GABA, glutamate and acetylcholine receptors and transporters. Stroke lesion data from two independent, matched cohorts comprising a total of 126 severely impaired acute stroke patients were used to assess NT-related network damage on a patient-specific basis across the distribution of each receptor and transporter. Multivariable logistic regression models were applied to evaluate the relationship between NT-informed network disconnections and functional outcomes 3 to 6 months post-stroke, operationalized by the modified Rankin scale. Analyses were adjusted for lesion-induced global network damage, age, sex, lesion volume and baseline neurological symptom burden. We present an innovative method for incorporating PET tracer data on various NT systems into normative structural connectome datasets. The resulting NT-informed connectivity maps revealed distinct spatial distributions consistent with the established literature. In both cohorts of severely impaired stroke patients, incorporating lesion-induced disruptions within specific NT systems provided more insights into variability in stroke outcomes than the structural disconnection data alone. Notably, greater damage to networks with high dopamine transporter density was associated with poorer functional recovery. Based on NT-informed structural connectivity maps with distinct topographical features for individual receptors and transporters, we show that lesion-induced disruptions in large-scale dopaminergic brain networks, beyond global structural network damage, may play a key role in stroke recovery. These insights hold significant translational potential for advancing personalized medicine in stroke care, such as targeted pharmacologic interventions.


  • Brain-wide spatial mapping of oscillatory activity during naturalistic motor behavior

    by Marlene Bönstrup, Tobias Schneider, Anne Bräuer, Jonas Ader, Arno Villringer, Joseph Classen

    We here provide a comprehensive account of brain rhythmic electric activity across spatial, spectral, and temporal scales, associated with ecologically valid, freely chosen, auditory cued, and visually guided movements of either hand. New and noteworthy, the brain-wide topography of movement preceding, short-lasting increases in low-frequency activity (3–5 Hz), recently identified as a promising target for on-demand neurostimulation in stroke rehabilitation, is described and compared with classically studied sensorimotor rhythms.


  • Open Access

    Cingulate and striatal hubs are linked to early skill learning

    by Hisato Sugata, Fumiaki Iwane, William Hayward, Valentina Azzollini, Debadatta Dash, Roberto F Salamanca-Giron, Marlene Bönstrup, Ethan R Buch, Leonardo G Cohen

    Early skill learning develops in the context of activity changes in distributed cortico-subcortical regions. Here, we investigated network hubs—centers of information integration and transmission—within the brain network supporting early skill learning. We recorded magnetoencephalographic (MEG) brain activity in healthy human subjects who learned a moderately difficult sequence skill with their non-dominant left hand. We then computed network hub strength by summing top 10% functional connectivity over 86 parcellated brain regions (AAL3 atlas) and five brain oscillatory frequency bands (alpha, low-, high-beta, low- and high-gamma). Virtually all skill gains developed during rest intervals of early learning (micro-offline gains). MEG hub strength in the alpha band (8-13Hz) in bilateral anterior cingulate (ACC) and caudate and in the low-beta band (13-16Hz) in bilateral caudate and right putamen correlated with micro-offline gains. These regions linked strongly with the hippocampus, parahippocampal cortex, and lingual and fusiform gyri. Thus, alpha and low-beta brain oscillatory activity in cingulate and striatal regions appear to contribute as hubs of information integration and transmission during early skill learning. Significance Statement Early learning of moderately difficult skill sequences develops over periods of rest interspersed with practice (micro-offline gains). We demonstrate here a link between alpha and low-beta oscillatory activity in a cingulate-hippocampo-striato network hubs and rest intervals of early learning.

  • Open Access

    Sequence action representations contextualize during early skill learning

    by Debadatta Dash, Fumiaki Iwane, William Hayward, Roberto Salamanca-Giron, Marlene Bonstrup, Ethan R. Buch, Leonardo G. Cohen

    Activities of daily living rely on our ability to acquire new motor skills composed of precise action sequences. Here, we asked if the millisecond-level neural representation of an action performed at different contextual sequence locations within a skill differentiates or remains stable during early motor learning. We first optimized machine learning decoders predictive of sequence-embedded finger movements from magnetoencephalographic (MEG) activity. Using this approach, we found that the neural representation of the same action performed in different contextual sequence locations, progressively differentiated—primarily during rest intervals of early learning (offline)—correlating with skill gains. In contrast, representational differentiation during practice (online) did not reflect learning. The regions contributing to this representational differentiation evolved with learning, shifting from the contralateral pre- and post-central cortex during early learning (trials 1–11) to increased involvement of the superior and middle frontal cortex once skill performance plateaued (trials 12–36). Thus, the neural substrates supporting finger movements and their representational differentiation during early skill learning differ from those supporting stable performance during the subsequent skill plateau period. Representational contextualization extended to Day 2, exhibiting specificity for the practiced skill sequence. Altogether, our findings indicate that sequence action representations contextually differentiate during early skill learning, an issue relevant to brain-computer interface applications in neurorehabilitation.

  • Open Access

    Generalization of procedural motor sequence learning after a single practice trial

    by B. P. Johnson, I. Iturrate, R. Y. Fakhreddine, M. Bönstrup, E. R. Buch, E. M. Robertson, L. G. Cohen

    When humans begin learning new motor skills, they typically display early rapid performance improvements. It is not well understood how knowledge acquired during this early skill learning period generalizes to new, related skills. Here, we addressed this question by investigating factors influencing generalization of early learning from a skill A to a different, but related skill B. Early skill generalization was tested over four experiments (N = 2095). Subjects successively learned two related motor sequence skills (skills A and B) over different practice schedules. Skill A and B sequences shared ordinal (i.e., matching keypress locations), transitional (i.e., ordered keypress pairs), parsing rule (i.e., distinct sequence events like repeated keypresses that can be used as a breakpoint for segmenting the sequence into smaller units) structures, or possessed no structure similarities. Results showed generalization for shared parsing rule structure between skills A and B after only a single 10-second practice trial of skill A. Manipulating the initial practice exposure to skill A (1 to 12 trials) and inter-practice rest interval (0–30 s) between skills A and B had no impact on parsing rule structure generalization. Furthermore, this generalization was not explained by stronger sensorimotor mapping between individual keypress actions and their symbolic representations. In contrast, learning from skill A did not generalize to skill B during early learning when the sequences shared only ordinal or transitional structure features. These results document sequence structure that can be very rapidly generalized during initial learning to facilitate generalization of skill.


  • Open Access

    Lesion aware automated processing pipeline for multimodal neuroimaging stroke data and The Virtual Brain (TVB)

    by Patrik Bey, Kiret Dhindsa, Amrit Kashyap, Michael Schirner, Jan Feldheim, Marlene Bönstrup, Robert Schulz, Bastian Cheng, Götz Thomalla, Christian Gerloff, et al.

    Background: Processing stroke magnetic resonance imaging (MRI) brain data can be susceptible to lesion-based abnormalities. In this study we developed and validated the Lesion Aware automated Processing Pipeline (LeAPP) that incorporates mitigation measures, improving volumetric and connectomics outputs compared to current standards in automated MRI processing pipelines. Methods: Building upon the Human Connectome Project (HCP) minimal processing pipeline, we introduced correction measures, such as cost-function masking and virtual brain transplant, and extended functional and diffusion processing to match acquisition protocols often found in a clinical context. A total of 51 participants (36 stroke patients (65.7±12.96 years, 18 female) and 15 healthy controls (69.2±7.4 years, 7 female)) were processed across four time points for patients (3-5, 30-40, 85-95, 340-380 days after stroke onset) and one time point for controls. Artificially lesioned brains (N=82), derived from healthy brains and informed by real stroke lesions were created, thus generating ground-truth data for validation. The processing pipeline and validation framework are available as containerized open-source software. Reconstruction quality has been quantified on whole brain level and for lesion affected and unaffected regions-of-interest (ROIs) using metrics like dice score, volume difference and center-of-gravity distance. Global and local level connectome reconstruction was assessed using node strength, node centrality and clustering coefficient. Results: The new pipeline LeAPP provides close reconstructions of the ground truth. Deviations in reconstructed averaged whole brain node strength and all ROI based volume and connectome metrics were significantly reduced compared to the HCP pipeline without stroke specific mitigation measures. Conclusions: LeAPP improves reconstruction quality of multimodal MRI processing for brain parcellation and structural connectome estimation significantly over the non-adapted HCP in the presence of lesions and provides a robust framework for diffusion and functional image processing of clinical stroke data. This novel open-source automated processing pipeline contributes to a development towards reproducible research. Graphical


  • Retracted
    Open Access

    Retracted and Replaced: Altered microstructure of the contralesional ventral premotor cortex and motor output after stroke

    by Paweł P Wróbel, Stephanie Guder, Jan F Feldheim, José A Graterol Pérez, Benedikt M Frey, Chi-Un Choe, Marlene Bönstrup, Bastian Cheng, Yogesh Rathi, Ofer Pasternak, et al.

    Cortical thickness (CT) analyses have provided valuable insights into changes in cortical brain structure after stroke and their association with recovery. Across studies though, relationships between cortical structure and function show inconsistent results. Recent developments in diffusion-weighted imaging of the cortex have paved the way to uncover hidden aspects of stroke-related alterations in cortical microstructure, going beyond CT as a surrogate for cortical macrostructure. We re-analysed clinical and imaging data of 42 well-recovered chronic stroke patients from two independent cohorts (mean age 64 years, 4 left-handed, 71% male, 16 right-sided strokes) and 33 healthy controls of similar age and gender. Cortical fractional anisotropy (FA) and CT values were obtained for six key sensorimotor areas of the contralesional hemisphere. The regions included the primary motor cortex, dorsal and ventral premotor cortex, supplementary and pre-supplementary motor areas, and primary somatosensory cortex. Linear models were estimated for group comparisons between patients and controls, and for correlations between cortical FA and CT and clinical scores. Compared to controls, stroke patients exhibited a reduction in FA in the contralesional ventral premotor cortex (P = 0.005). FA of the other regions, and CT did not show a comparable group difference. Higher FA of the ventral premotor cortex, but not CT, was positively associated with residual grip force in the stroke patients. These data provide novel evidence that the contralesional ventral premotor cortex might constitute a key sensorimotor area particularly susceptible to stroke-related alterations in cortical microstructure as measured by diffusion MRI and they suggest a link between these changes and residual motor output after stroke.