Esther E. Bron
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
Research Subjects
Esther E. Bron has a total of 500 co-authors. Here are the top co-authors by number of shared publications:
Stefan Klein
Department of Radiology and Nuclear Medicine, Erasmus MC
22 shared publicationsWiro J. Niessen
Department of Radiology and Nuclear Medicine, Erasmus MC
11 shared publications- Wiro J. Niessen
Erasmus MC
11 shared publications - Rebecca M. E. Steketee
Department of Radiology and Nuclear Medicine, Erasmus MC University Medical Center, Rotterdam, The Netherlands
10 shared publications Marion Smits
Erasmus MC
9 shared publications- Geert Jan Biessels
Department of Neurology, University Medical Center Utrecht, Utrecht, The Netherlands
8 shared publications
Latest Publications
- Open Access
- 23 October 2026
- Published by Frontiers Media SA in Frontiers in Applied Mathematics and Statistics
- Vol. 12, 1923866
Federated deep learning for privacy-preserving cardiovascular disease risk prediction
Cardiovascular disease risk prediction models often rely on data from a single institution or centrally pooled datasets. Extending these models across institutions could be limited by privacy regulations and constraints on sharing patient-level data. Federated learning enables collaborative model development without transferring sensitive patient data, but its application in healthcare remains challenging because datasets often differ in size, population characteristics, and outcome definitions. In this study, we present a federated deep learning approach for privacy-preserving cardiovascular disease risk prediction that integrates two population-based cohorts with different characteristics: Lifelines, including 148,230 participants meeting the study inclusion criteria with self-reported outcomes, and the Rotterdam Study, including a smaller cohort of 10,155 participants with digitally linked clinical outcomes. Model performance was primarily evaluated on the Rotterdam Study because of its complete follow-up. Deep survival models trained using federated learning achieved a higher mean C-statistic than models trained locally without federation. For the Rotterdam Study, the C-statistic increased from 0.723 (95% CI: 0.714–0.732) to 0.737 (95% CI: 0.727–0.747). For Lifelines, the C-statistic increased from 0.791 (95% CI: 0.783–0.799) to 0.795 (95% CI: 0.785–0.804). These findings suggest that federated deep learning across heterogeneous cohorts can improve the discrimination of cardiovascular disease risk prediction models while keeping individual-level patient data within the participating institutions.
- 23 October 2026
- Published by Frontiers Media SA in Frontiers in Applied Mathematics and Statistics
- Vol. 12, 1923866
Open AccessFederated deep learning for privacy-preserving cardiovascular disease risk prediction
Cardiovascular disease risk prediction models often rely on data from a single institution or centrally pooled datasets. Extending these models across institutions could be limited by privacy regulations and constraints on sharing patient-level data. Federated learning enables collaborative model development without transferring sensitive patient data, but its application in healthcare remains challenging because datasets often differ in size, population characteristics, and outcome definitions. In this study, we present a federated deep learning approach for privacy-preserving cardiovascular disease risk prediction that integrates two population-based cohorts with different characteristics: Lifelines, including 148,230 participants meeting the study inclusion criteria with self-reported outcomes, and the Rotterdam Study, including a smaller cohort of 10,155 participants with digitally linked clinical outcomes. Model performance was primarily evaluated on the Rotterdam Study because of its complete follow-up. Deep survival models trained using federated learning achieved a higher mean C-statistic than models trained locally without federation. For the Rotterdam Study, the C-statistic increased from 0.723 (95% CI: 0.714–0.732) to 0.737 (95% CI: 0.727–0.747). For Lifelines, the C-statistic increased from 0.791 (95% CI: 0.783–0.799) to 0.795 (95% CI: 0.785–0.804). These findings suggest that federated deep learning across heterogeneous cohorts can improve the discrimination of cardiovascular disease risk prediction models while keeping individual-level patient data within the participating institutions.
- Open Access
- 3 October 2026
- Published by Elsevier in Cerebral Circulation - Cognition and Behavior
- Vol. 11
Air pollution, neighbourhood walkability and cerebrovascular small vessel disease in patients with cardiovascular disorders along the heart-brain axis
- 3 October 2026
- Published by Elsevier in Cerebral Circulation - Cognition and Behavior
- Vol. 11
Open AccessAir pollution, neighbourhood walkability and cerebrovascular small vessel disease in patients with cardiovascular disorders along the heart-brain axis
- Open Access
- 1 September 2026
- Published by Wolters Kluwer Health in Neurology Open Access
- Vol. 2 (3) , e000120
Atrophy Patterns Across Dementia Etiologies Assessed by MRI-Based Clustering of a Memory Clinic Population
Background and ObjectivesIn clinical practice, it is difficult to accurately diagnose dementia etiologies, although brain MRI has strong diagnostic potential. Unsupervised data-driven methods can p...
- 1 September 2026
- Published by Wolters Kluwer Health in Neurology Open Access
- Vol. 2 (3) , e000120
Open AccessAtrophy Patterns Across Dementia Etiologies Assessed by MRI-Based Clustering of a Memory Clinic Population
Background and ObjectivesIn clinical practice, it is difficult to accurately diagnose dementia etiologies, although brain MRI has strong diagnostic potential. Unsupervised data-driven methods can p...
- Open Access
- 9 October 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 198 (Pt A) , 111180
MyDigiTwin: A privacy-preserving framework for personalized cardiovascular risk prediction and scenario exploration
- 9 October 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 198 (Pt A) , 111180
Open AccessMyDigiTwin: A privacy-preserving framework for personalized cardiovascular risk prediction and scenario exploration
- Open Access
- 30 July 2025
- Published by Springer Nature in Journal of Healthcare Informatics Research
- Vol. 9 (4) , 606-628
Multi-Cohort Federated Learning Shows Synergy in Mortality Prediction for MRI-Based and Metabolomics-Based Age Scores
While biological age scores have been shown to characterize aging by estimating chronological age based on physiological biomarkers, interactions between different age scores are largely unknown. To study this, large-scale multi-modal data are crucial. However, such data are scarce as population-based cohorts are generally restricted in sharing their data. Here, we employ federated learning to study the relationship between the two types of biological age scores: BrainAge based on brain MRI and MetaboAge based on metabolites. Using three large population-based cohorts, we trained a federated deep learning model to estimate BrainAge and compared its performance to models trained in a single cohort. The federated BrainAge model yielded significantly lower error for age prediction across the cohorts than locally trained models. Harmonizing the age interval between cohorts further improved BrainAge accuracy. Subsequently, we compared BrainAge and MetaboAge by performing association analysis and survival analysis for dementia and mortality prediction to further characterize both scores. The association analysis showed a weak association between BrainAge and MetaboAge, while the survival analysis indicated complementary predictive values for the mortality risk of the two scores. Federated learning has been shown to be a valuable technique for enabling the use of research cohorts that are restricted in data sharing. We conclude that BrainAge and MetaboAge act synergetically for the prediction of time to all-cause mortality, and both aging scores capture different aspects of the aging process.
- 30 July 2025
- Published by Springer Nature in Journal of Healthcare Informatics Research
- Vol. 9 (4) , 606-628
Open AccessMulti-Cohort Federated Learning Shows Synergy in Mortality Prediction for MRI-Based and Metabolomics-Based Age Scores
While biological age scores have been shown to characterize aging by estimating chronological age based on physiological biomarkers, interactions between different age scores are largely unknown. To study this, large-scale multi-modal data are crucial. However, such data are scarce as population-based cohorts are generally restricted in sharing their data. Here, we employ federated learning to study the relationship between the two types of biological age scores: BrainAge based on brain MRI and MetaboAge based on metabolites. Using three large population-based cohorts, we trained a federated deep learning model to estimate BrainAge and compared its performance to models trained in a single cohort. The federated BrainAge model yielded significantly lower error for age prediction across the cohorts than locally trained models. Harmonizing the age interval between cohorts further improved BrainAge accuracy. Subsequently, we compared BrainAge and MetaboAge by performing association analysis and survival analysis for dementia and mortality prediction to further characterize both scores. The association analysis showed a weak association between BrainAge and MetaboAge, while the survival analysis indicated complementary predictive values for the mortality risk of the two scores. Federated learning has been shown to be a valuable technique for enabling the use of research cohorts that are restricted in data sharing. We conclude that BrainAge and MetaboAge act synergetically for the prediction of time to all-cause mortality, and both aging scores capture different aspects of the aging process.
- 29 April 2025
- Published by Springer Nature
- p. 122-131
Coronary Artery Calcium Scoring from Non-contrast Cardiac CT Using Deep Learning with External Validation
- 29 April 2025
- Published by Springer Nature
- p. 122-131
Coronary Artery Calcium Scoring from Non-contrast Cardiac CT Using Deep Learning with External Validation
- Open Access
- 17 April 2025
- Published by SAGE Publications in Journal of Alzheimer’s Disease
- Vol. 105 (2) , 519-530
Dynamic PET imaging in patients with unilateral carotid occlusion shows lateralized cerebral hypoperfusion, but no amyloid binding
Background: Carotid occlusive disease is a risk factor for cognitive decline. A possible underlying etiology is that hemodynamic impairment results in decreased cerebral perfusion, exacerbated amyloid-β accumulation (Aβ) and poorer cognitive performance. Objective: We aimed to determine whether patients with unilateral internal carotid artery (ICA) occlusion have less cerebral perfusion and more Aβ in the ipsilateral than in the contralateral hemisphere, and whether perfusion and Aβ are associated with cognitive functioning. Methods: We included 20 patients (age 67.2 ± 7.0 years, 8 females, MMSE 29 [27–29]) with unilateral ICA occlusion, which underwent neuropsychological assessment and dynamic 18 F-Florbetaben positron emission tomography (PET). Global and regional relative perfusion (R 1 ) and binding potential (BP ND ) were obtained from the PET-images using a simplified reference tissue model. We performed Wilcoxon signed-rank tests to examine differences between hemispheres within subjects and linear regression to investigate associations with cognitive functioning. Results: Median global R 1 was 0.911 (0.883–0.950) and global BP ND was 0.172 (0.129–0.187). R 1 was lower in the hemisphere ipsilateral to the ICA occlusion than in the contralateral hemisphere (0.899 [0.876–0.921] versus 0.935 [0.889–0.970]). BP ND did not differ significantly between hemispheres (ipsilateral 0.172 [0.124–0.181] versus contralateral 0.168 [0.137–0.191]). Neither cerebral perfusion nor Aβ burden were associated with cognitive functioning. Conclusions: Patients with unilateral ICA occlusion did not have more Aβ in the ipsilateral hemisphere than in the contralateral hemisphere despite ipsilateral hypoperfusion. Perfusion and Aβ were unrelated to cognitive functioning. This indicates that cognitive impairment in patients with ICA occlusion is not due to exacerbated Aβ accumulation.
- 17 April 2025
- Published by SAGE Publications in Journal of Alzheimer’s Disease
- Vol. 105 (2) , 519-530
Open AccessDynamic PET imaging in patients with unilateral carotid occlusion shows lateralized cerebral hypoperfusion, but no amyloid binding
Background: Carotid occlusive disease is a risk factor for cognitive decline. A possible underlying etiology is that hemodynamic impairment results in decreased cerebral perfusion, exacerbated amyloid-β accumulation (Aβ) and poorer cognitive performance. Objective: We aimed to determine whether patients with unilateral internal carotid artery (ICA) occlusion have less cerebral perfusion and more Aβ in the ipsilateral than in the contralateral hemisphere, and whether perfusion and Aβ are associated with cognitive functioning. Methods: We included 20 patients (age 67.2 ± 7.0 years, 8 females, MMSE 29 [27–29]) with unilateral ICA occlusion, which underwent neuropsychological assessment and dynamic 18 F-Florbetaben positron emission tomography (PET). Global and regional relative perfusion (R 1 ) and binding potential (BP ND ) were obtained from the PET-images using a simplified reference tissue model. We performed Wilcoxon signed-rank tests to examine differences between hemispheres within subjects and linear regression to investigate associations with cognitive functioning. Results: Median global R 1 was 0.911 (0.883–0.950) and global BP ND was 0.172 (0.129–0.187). R 1 was lower in the hemisphere ipsilateral to the ICA occlusion than in the contralateral hemisphere (0.899 [0.876–0.921] versus 0.935 [0.889–0.970]). BP ND did not differ significantly between hemispheres (ipsilateral 0.172 [0.124–0.181] versus contralateral 0.168 [0.137–0.191]). Neither cerebral perfusion nor Aβ burden were associated with cognitive functioning. Conclusions: Patients with unilateral ICA occlusion did not have more Aβ in the ipsilateral hemisphere than in the contralateral hemisphere despite ipsilateral hypoperfusion. Perfusion and Aβ were unrelated to cognitive functioning. This indicates that cognitive impairment in patients with ICA occlusion is not due to exacerbated Aβ accumulation.
- Open Access
- 6 March 2025
- Published by Elsevier in Medical Image Analysis
- Vol. 102, 103529
AI-based association analysis for medical imaging using latent-space geometric confounder correction
- 6 March 2025
- Published by Elsevier in Medical Image Analysis
- Vol. 102, 103529
Open AccessAI-based association analysis for medical imaging using latent-space geometric confounder correction
- Open Access
- 14 February 2025
- Published by Elsevier in eClinicalMedicine
- Vol. 81, 103092
Cerebral blood flow and cognitive functioning in patients undergoing transcatheter aortic valve implantation
- 14 February 2025
- Published by Elsevier in eClinicalMedicine
- Vol. 81, 103092
Open AccessCerebral blood flow and cognitive functioning in patients undergoing transcatheter aortic valve implantation
- 1 December 2024
- Published by Wolters Kluwer Health in Stroke
- Vol. 55 (12) , 2863-2871
Normative Population–Derived Data for MRI Manifestations of Cerebral Small Vessel Disease
BACKGROUND: Cerebral small vessel disease (SVD) is manifested on magnetic resonance imaging (MRI) by white matter hyperintensities, lacunes, microbleeds, and atrophy. While these manifestations can be part of normal aging, a high burden has been associated with cognitive impairment and vascular events. Distinguishing between normal versus abnormal SVD lesion burden in clinical practice remains complex. Our objective is to establish age- and sex-specific normative data for MRI manifestations of SVD, to support clinical assessment in individual patients. METHODS: We used 11 465 MRI scans from 5402 participants of the Rotterdam Study, the Netherlands, an ongoing prospective population-based cohort since 1990, to develop percentile curves for white matter hyperintensities and brain parenchymal fraction and probability curves for the prevalence and count of lacunes and microbleeds, across ages 45 to 100 years, stratified by sex. RESULTS: Participants were primarily White (≈97%), with a mean age at first scan of 64.7 (range, 45.7–97.9) years, and 55.7% being female participants. For all SVD MRI manifestations, the curves demonstrated nonlinear trends, with accelerating burden with advancing age (eg, doubling of white matter hyperintensity fraction every 10 years). Regarding brain parenchymal fraction, a decline was seen earlier in male participants (≈45 years) than female participants (≈60 years) and was more pronounced in male participants over time. Female participants had slightly higher white matter hyperintensity fractions compared with male participants across all ages. Lacunes and microbleeds were more frequently found in male participants than in female participants, and microbleeds were more prevalent than lacunes. CONCLUSIONS: We provide comprehensive normative data for different MRI manifestations of SVD, presented as percentile and probability curves by age, stratified by sex. This can aid clinicians to actually quantify the SVD burden on an individual patient’s MRI scans and detect patterns of abnormality.
- 1 December 2024
- Published by Wolters Kluwer Health in Stroke
- Vol. 55 (12) , 2863-2871
Normative Population–Derived Data for MRI Manifestations of Cerebral Small Vessel Disease
BACKGROUND: Cerebral small vessel disease (SVD) is manifested on magnetic resonance imaging (MRI) by white matter hyperintensities, lacunes, microbleeds, and atrophy. While these manifestations can be part of normal aging, a high burden has been associated with cognitive impairment and vascular events. Distinguishing between normal versus abnormal SVD lesion burden in clinical practice remains complex. Our objective is to establish age- and sex-specific normative data for MRI manifestations of SVD, to support clinical assessment in individual patients. METHODS: We used 11 465 MRI scans from 5402 participants of the Rotterdam Study, the Netherlands, an ongoing prospective population-based cohort since 1990, to develop percentile curves for white matter hyperintensities and brain parenchymal fraction and probability curves for the prevalence and count of lacunes and microbleeds, across ages 45 to 100 years, stratified by sex. RESULTS: Participants were primarily White (≈97%), with a mean age at first scan of 64.7 (range, 45.7–97.9) years, and 55.7% being female participants. For all SVD MRI manifestations, the curves demonstrated nonlinear trends, with accelerating burden with advancing age (eg, doubling of white matter hyperintensity fraction every 10 years). Regarding brain parenchymal fraction, a decline was seen earlier in male participants (≈45 years) than female participants (≈60 years) and was more pronounced in male participants over time. Female participants had slightly higher white matter hyperintensity fractions compared with male participants across all ages. Lacunes and microbleeds were more frequently found in male participants than in female participants, and microbleeds were more prevalent than lacunes. CONCLUSIONS: We provide comprehensive normative data for different MRI manifestations of SVD, presented as percentile and probability curves by age, stratified by sex. This can aid clinicians to actually quantify the SVD burden on an individual patient’s MRI scans and detect patterns of abnormality.