Michiel Kallenberg
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Michiel Kallenberg has a total of 64 co-authors. Here are the top co-authors by number of shared publications:
- Ioannis N. Athanasiadis
Wageningen University and Research
6 shared publications - Hilmy Baja
Wageningen University and Research, Artificial Intelligence, PO Box 16, Wageningen, 6700 AA, the Netherlands
4 shared publications Ron van Bree
4 shared publications- Aike Potze
Wageningen University and Research, Artificial Intelligence, PO Box 16, Wageningen, 6700 AA, the Netherlands
2 shared publications - Alex Ruane
NASA Goddard Institute for Space Studies, New York, USA
2 shared publications - Amit Kumar Srivastava
Leibniz Centre for Agricultural Landscape Research, Simulation and Data Science, Eberswalder Straße 84, Müncheberg, 15374, Germany
2 shared publications
Latest Publications
- Open Access
- 8 August 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
MRICombo: a deep-learning-based framework for universal volumetric segmentation grading-staging and malignancy detection across heterogeneous MRI
Comprehensive magnetic resonance imaging (MRI) analysis in oncology involves multiple interrelated tasks including volumetric segmentation, grading, staging, and malignancy detection. However, most existing deep learning models are task-specific or sequence-specific, lacking the generalizability required for heterogeneous sequences. Here we present MRICombo, a unified multi-expert deep learning framework for universal anatomical delineation and tumor characterization across 9 heterogeneous imaging sequences. Developed using 7,380 MRI sequences from 2,354 individuals, MRICombo achieves state-of-the-art performance with mean Dice similarity coefficients of 0.836 for segmenting 14 critical anatomical structures and 0.625 for labeling 11 major tumor types. It also attains a mean area under the receiver operating characteristic curve (AUROC) of 0.920 for glioma grading, bladder and nasopharyngeal cancer staging, and breast and liver tumor malignancy detection. External validation on four independent datasets (1082 sequences from 734 individuals) and transfer learning evaluation (512 individuals) confirm robust cross-protocol generalizability. Furthermore, MRICombo supports flexible inference with missing sequences and offers decision interpretability through sequence clustering and expert contribution analysis. As a unified clinical solution, MRICombo significantly reduces deployment costs and has the potential to streamline diagnostic workflows, supporting more personalized oncology care.
- 8 August 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessMRICombo: a deep-learning-based framework for universal volumetric segmentation grading-staging and malignancy detection across heterogeneous MRI
Comprehensive magnetic resonance imaging (MRI) analysis in oncology involves multiple interrelated tasks including volumetric segmentation, grading, staging, and malignancy detection. However, most existing deep learning models are task-specific or sequence-specific, lacking the generalizability required for heterogeneous sequences. Here we present MRICombo, a unified multi-expert deep learning framework for universal anatomical delineation and tumor characterization across 9 heterogeneous imaging sequences. Developed using 7,380 MRI sequences from 2,354 individuals, MRICombo achieves state-of-the-art performance with mean Dice similarity coefficients of 0.836 for segmenting 14 critical anatomical structures and 0.625 for labeling 11 major tumor types. It also attains a mean area under the receiver operating characteristic curve (AUROC) of 0.920 for glioma grading, bladder and nasopharyngeal cancer staging, and breast and liver tumor malignancy detection. External validation on four independent datasets (1082 sequences from 734 individuals) and transfer learning evaluation (512 individuals) confirm robust cross-protocol generalizability. Furthermore, MRICombo supports flexible inference with missing sequences and offers decision interpretability through sequence clustering and expert contribution analysis. As a unified clinical solution, MRICombo significantly reduces deployment costs and has the potential to streamline diagnostic workflows, supporting more personalized oncology care.
- Open Access
- 11 June 2026
- Published by Copernicus GmbH in Earth System Science Data
- Vol. 18 (6) , 3997-4018
CY-Bench: a comprehensive benchmark dataset for sub-national crop yield forecasting
In-season, pre-harvest crop yield forecasts are essential for enhancing transparency in commodity markets and improving food security. They play a key role in increasing resilience to climate change and extreme events and thus contribute to the United Nations’ Sustainable Development Goal 2 of zero hunger. Pre-harvest crop yield forecasting is a complex task, as several interacting factors contribute to yield formation, including in-season weather variability, extreme events, long-term climate change, soil, pests, diseases and farm management decisions. Several modeling approaches have been employed to capture complex interactions among such predictors and crop yields. Prior research for in-season, pre-harvest crop yield forecasting has primarily been case-study based, which makes it difficult to compare modeling approaches and measure progress systematically. To address this gap, we introduce CY-Bench (Crop Yield Benchmark), a comprehensive dataset and benchmark to forecast maize and wheat yields at a global scale. CY-Bench was conceptualized and developed within the Machine Learning team of the Agricultural Model Intercomparison and Improvement Project (AgML) in collaboration with agronomists, climate scientists, and machine learning researchers. It features publicly available sub-national yield statistics and relevant predictors, such as weather data, soil characteristics, and remote sensing indicators, that have been pre-processed, standardized, and harmonized across spatio-temporal scales. With CY-Bench, we aim to: (i) establish a standardized framework for developing and evaluating data-driven models across diverse farming systems in more than 25 countries across six continents; (ii) enable robust and reproducible model comparisons that address real-world operational challenges; (iii) provide an openly accessible dataset to the earth system science and machine learning communities, facilitating research on time series forecasting, domain adaptation, and online learning. The dataset (https://doi.org/10.5281/zenodo.11502142, Kallenberg et al., 2025) and accompanying code (https://doi.org/10.5281/zenodo.20456375, Kallenberg et al., 2026) are openly available to support the continuous development of advanced data driven models for crop yield forecasting to enhance decision-making on food security.
- 11 June 2026
- Published by Copernicus GmbH in Earth System Science Data
- Vol. 18 (6) , 3997-4018
Open AccessCY-Bench: a comprehensive benchmark dataset for sub-national crop yield forecasting
In-season, pre-harvest crop yield forecasts are essential for enhancing transparency in commodity markets and improving food security. They play a key role in increasing resilience to climate change and extreme events and thus contribute to the United Nations’ Sustainable Development Goal 2 of zero hunger. Pre-harvest crop yield forecasting is a complex task, as several interacting factors contribute to yield formation, including in-season weather variability, extreme events, long-term climate change, soil, pests, diseases and farm management decisions. Several modeling approaches have been employed to capture complex interactions among such predictors and crop yields. Prior research for in-season, pre-harvest crop yield forecasting has primarily been case-study based, which makes it difficult to compare modeling approaches and measure progress systematically. To address this gap, we introduce CY-Bench (Crop Yield Benchmark), a comprehensive dataset and benchmark to forecast maize and wheat yields at a global scale. CY-Bench was conceptualized and developed within the Machine Learning team of the Agricultural Model Intercomparison and Improvement Project (AgML) in collaboration with agronomists, climate scientists, and machine learning researchers. It features publicly available sub-national yield statistics and relevant predictors, such as weather data, soil characteristics, and remote sensing indicators, that have been pre-processed, standardized, and harmonized across spatio-temporal scales. With CY-Bench, we aim to: (i) establish a standardized framework for developing and evaluating data-driven models across diverse farming systems in more than 25 countries across six continents; (ii) enable robust and reproducible model comparisons that address real-world operational challenges; (iii) provide an openly accessible dataset to the earth system science and machine learning communities, facilitating research on time series forecasting, domain adaptation, and online learning. The dataset (https://doi.org/10.5281/zenodo.11502142, Kallenberg et al., 2025) and accompanying code (https://doi.org/10.5281/zenodo.20456375, Kallenberg et al., 2026) are openly available to support the continuous development of advanced data driven models for crop yield forecasting to enhance decision-making on food security.
- Open Access
- 25 November 2025
- Published by Springer Nature in Research Square
MRICombo: a deep-learning-based framework for universal volume segmentation, grading-staging, and malignancy detection across heterogeneous MRI
Comprehensive magnetic resonance imaging (MRI) analysis in oncology involves multiple interrelated tasks including volumetric segmentation, grading, staging, and malignancy detection. However, most existing deep learning models are task-specific or sequence-specific, lacking the generalizability required for heterogeneous sequences. Here we present MRICombo, a unified multi-expert deep learning framework for universal anatomical delineation and tumor characterization across 9 heterogeneous imaging sequences. Developed using 7,376 MRI sequences from 2,282 individuals, MRICombo achieves state-of-the-art performance with mean Dice similarity coefficients of 0.839 for segmenting 14 critical anatomical structures and 0.633 for labeling 11 major tumor types. It also attains a mean area under the receiver operating characteristic curve (AUROC) of 0.913 for glioma grading, bladder and nasopharyngeal cancer staging, and breast and liver tumor malignancy detection. External validation on four independent datasets (998 sequences from 734 individuals) and transfer learning evaluation (260 individuals) confirm robust cross-protocols generalizability. Furthermore, MRICombo supports flexible inference with missing sequences and offers decision interpretability through sequence clustering and expert contribution analysis. As an all-in-one clinical solution, MRICombo significantly reduces deployment costs and streamlines diagnostic workflows, supporting more personalized oncology care.
- 25 November 2025
- Published by Springer Nature in Research Square
Open AccessMRICombo: a deep-learning-based framework for universal volume segmentation, grading-staging, and malignancy detection across heterogeneous MRI
Comprehensive magnetic resonance imaging (MRI) analysis in oncology involves multiple interrelated tasks including volumetric segmentation, grading, staging, and malignancy detection. However, most existing deep learning models are task-specific or sequence-specific, lacking the generalizability required for heterogeneous sequences. Here we present MRICombo, a unified multi-expert deep learning framework for universal anatomical delineation and tumor characterization across 9 heterogeneous imaging sequences. Developed using 7,376 MRI sequences from 2,282 individuals, MRICombo achieves state-of-the-art performance with mean Dice similarity coefficients of 0.839 for segmenting 14 critical anatomical structures and 0.633 for labeling 11 major tumor types. It also attains a mean area under the receiver operating characteristic curve (AUROC) of 0.913 for glioma grading, bladder and nasopharyngeal cancer staging, and breast and liver tumor malignancy detection. External validation on four independent datasets (998 sequences from 734 individuals) and transfer learning evaluation (260 individuals) confirm robust cross-protocols generalizability. Furthermore, MRICombo supports flexible inference with missing sequences and offers decision interpretability through sequence clustering and expert contribution analysis. As an all-in-one clinical solution, MRICombo significantly reduces deployment costs and streamlines diagnostic workflows, supporting more personalized oncology care.
- Open Access
- 11 September 2025
- Published by Elsevier in Smart Agricultural Technology
- Vol. 12
Interoperable agricultural digital twins with reinforcement learning intelligence
- 11 September 2025
- Published by Elsevier in Smart Agricultural Technology
- Vol. 12
Open AccessInteroperable agricultural digital twins with reinforcement learning intelligence
- Open Access
- 4 June 2025
- Published by Elsevier in Computers and Electronics in Agriculture
- Vol. 237
Adaptive fertilizer management for optimizing nitrogen use efficiency with constrained reinforcement learning
- 4 June 2025
- Published by Elsevier in Computers and Electronics in Agriculture
- Vol. 237
Open AccessAdaptive fertilizer management for optimizing nitrogen use efficiency with constrained reinforcement learning
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (27) , 27831-27840
To Measure or Not: A Cost-Sensitive, Selective Measuring Environment for Agricultural Management Decisions with Reinforcement Learning
Farmers rely on in-field observations to make well-informed crop management decisions to maximize profit and minimize adverse environmental impact. However, obtaining real-world crop state measurements is labor-intensive, time-consuming and expensive. In most cases, it is not feasible to gather crop state measurements before every decision moment. Moreover, in previous research pertaining to farm management optimization, these observations are often assumed to be readily available without any cost, which is unrealistic. Hence, enabling optimization without the need to have *temporally complete* crop state observations is important. An approach to that problem is to include measuring as part of decision making. As a solution, we apply reinforcement learning (RL) to recommend opportune moments to simultaneously measure crop features and apply nitrogen fertilizer. With realistic considerations, we design an RL environment with explicit crop feature measuring costs. While balancing costs, we find that an RL agent, trained with recurrent PPO, discovers adaptive measuring policies that follow critical crop development stages, with results aligned by what domain experts would consider a sensible approach. Our results highlight the importance of measuring when crop feature measurements are not readily available.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (27) , 27831-27840
Open AccessTo Measure or Not: A Cost-Sensitive, Selective Measuring Environment for Agricultural Management Decisions with Reinforcement Learning
Farmers rely on in-field observations to make well-informed crop management decisions to maximize profit and minimize adverse environmental impact. However, obtaining real-world crop state measurements is labor-intensive, time-consuming and expensive. In most cases, it is not feasible to gather crop state measurements before every decision moment. Moreover, in previous research pertaining to farm management optimization, these observations are often assumed to be readily available without any cost, which is unrealistic. Hence, enabling optimization without the need to have *temporally complete* crop state observations is important. An approach to that problem is to include measuring as part of decision making. As a solution, we apply reinforcement learning (RL) to recommend opportune moments to simultaneously measure crop features and apply nitrogen fertilizer. With realistic considerations, we design an RL environment with explicit crop feature measuring costs. While balancing costs, we find that an RL agent, trained with recurrent PPO, discovers adaptive measuring policies that follow critical crop development stages, with results aligned by what domain experts would consider a sensible approach. Our results highlight the importance of measuring when crop feature measurements are not readily available.
- Open Access
- 12 March 2025
- Published by Copernicus GmbH in EGUsphere
- Vol. 2025, 1-28
CY-Bench: A comprehensive benchmark dataset for sub-national crop yield forecasting
In-season, pre-harvest crop yield forecasts are essential for enhancing transparency in commodity markets and improving food security. They play a key role in increasing resilience to climate change and extreme events and thus contribute to the United Nations’ Sustainable Development Goal 2 of zero hunger. Pre-harvest crop yield forecasting is a complex task, as several interacting factors contribute to yield formation, including in-season weather variability, extreme events, long-term climate change, soil, pests, diseases and farm management decisions. Several modeling approaches have been employed to capture complex interactions among such predictors and crop yields. Prior research for in-season, pre-harvest crop yield forecasting has primarily been case-study based, which makes it difficult to compare modeling approaches and measure progress systematically. To address this gap, we introduce CY-Bench (Crop Yield Benchmark), a comprehensive dataset and benchmark to forecast maize and wheat yields at a global scale. CY-Bench was conceptualized and developed within the Machine Learning team of the Agricultural Model Intercomparison and Improvement Project (AgML) in collaboration with agronomists, climate scientists, and machine learning researchers. It features publicly available sub-national yield statistics and relevant predictors—such as weather data, soil characteristics, and remote sensing indicators—that have been pre-processed, standardized, and harmonized across spatio-temporal scales. With CY-Bench, we aim to: (i) establish a standardized framework for developing and evaluating data-driven models across diverse farming systems in more than 25 countries across six continents; (ii) enable robust and reproducible model comparisons that address real-world operational challenges; (iii) provide an openly accessible dataset to the earth system science and machine learning communities, facilitating research on time series forecasting, domain adaptation, and online learning. The dataset (https://doi.org/10.5281/zenodo.11502142, (Paudel et al., 2025a)) and accompanying code (https://github.com/WUR-AI/AgML-CY-Bench, (Paudel et al., 2025b))) are openly available to support the continuous development of advanced data driven models for crop yield forecasting to enhance decision-making on food security.
- 12 March 2025
- Published by Copernicus GmbH in EGUsphere
- Vol. 2025, 1-28
Open AccessCY-Bench: A comprehensive benchmark dataset for sub-national crop yield forecasting
In-season, pre-harvest crop yield forecasts are essential for enhancing transparency in commodity markets and improving food security. They play a key role in increasing resilience to climate change and extreme events and thus contribute to the United Nations’ Sustainable Development Goal 2 of zero hunger. Pre-harvest crop yield forecasting is a complex task, as several interacting factors contribute to yield formation, including in-season weather variability, extreme events, long-term climate change, soil, pests, diseases and farm management decisions. Several modeling approaches have been employed to capture complex interactions among such predictors and crop yields. Prior research for in-season, pre-harvest crop yield forecasting has primarily been case-study based, which makes it difficult to compare modeling approaches and measure progress systematically. To address this gap, we introduce CY-Bench (Crop Yield Benchmark), a comprehensive dataset and benchmark to forecast maize and wheat yields at a global scale. CY-Bench was conceptualized and developed within the Machine Learning team of the Agricultural Model Intercomparison and Improvement Project (AgML) in collaboration with agronomists, climate scientists, and machine learning researchers. It features publicly available sub-national yield statistics and relevant predictors—such as weather data, soil characteristics, and remote sensing indicators—that have been pre-processed, standardized, and harmonized across spatio-temporal scales. With CY-Bench, we aim to: (i) establish a standardized framework for developing and evaluating data-driven models across diverse farming systems in more than 25 countries across six continents; (ii) enable robust and reproducible model comparisons that address real-world operational challenges; (iii) provide an openly accessible dataset to the earth system science and machine learning communities, facilitating research on time series forecasting, domain adaptation, and online learning. The dataset (https://doi.org/10.5281/zenodo.11502142, (Paudel et al., 2025a)) and accompanying code (https://github.com/WUR-AI/AgML-CY-Bench, (Paudel et al., 2025b))) are openly available to support the continuous development of advanced data driven models for crop yield forecasting to enhance decision-making on food security.
- Open Access
To Measure or Not: A Cost-Sensitive, Selective Measuring Environment for Agricultural Management Decisions with Reinforcement Learning
Farmers rely on in-field observations to make well-informed crop management decisions to maximize profit and minimize adverse environmental impact. However, obtaining real-world crop state measurements is labor-intensive, time-consuming and expensive. In most cases, it is not feasible to gather crop state measurements before every decision moment. Moreover, in previous research pertaining to farm management optimization, these observations are often assumed to be readily available without any cost, which is unrealistic. Hence, enabling optimization without the need to have temporally complete crop state observations is important. An approach to that problem is to include measuring as part of decision making. As a solution, we apply reinforcement learning (RL) to recommend opportune moments to simultaneously measure crop features and apply nitrogen fertilizer. With realistic considerations, we design an RL environment with explicit crop feature measuring costs. While balancing costs, we find that an RL agent, trained with recurrent PPO, discovers adaptive measuring policies that follow critical crop development stages, with results aligned by what domain experts would consider a sensible approach. Our results highlight the importance of measuring when crop feature measurements are not readily available.
Open AccessTo Measure or Not: A Cost-Sensitive, Selective Measuring Environment for Agricultural Management Decisions with Reinforcement Learning
Farmers rely on in-field observations to make well-informed crop management decisions to maximize profit and minimize adverse environmental impact. However, obtaining real-world crop state measurements is labor-intensive, time-consuming and expensive. In most cases, it is not feasible to gather crop state measurements before every decision moment. Moreover, in previous research pertaining to farm management optimization, these observations are often assumed to be readily available without any cost, which is unrealistic. Hence, enabling optimization without the need to have temporally complete crop state observations is important. An approach to that problem is to include measuring as part of decision making. As a solution, we apply reinforcement learning (RL) to recommend opportune moments to simultaneously measure crop features and apply nitrogen fertilizer. With realistic considerations, we design an RL environment with explicit crop feature measuring costs. While balancing costs, we find that an RL agent, trained with recurrent PPO, discovers adaptive measuring policies that follow critical crop development stages, with results aligned by what domain experts would consider a sensible approach. Our results highlight the importance of measuring when crop feature measurements are not readily available.
- Open Access
- 25 September 2023
- Published by Cambridge University Press (CUP) in Environmental Data Science
- Vol. 2, e34
Nitrogen management with reinforcement learning and crop growth models
The growing need for agricultural products and the challenges posed by environmental and economic factors have created a demand for enhanced agricultural systems management. Machine learning has increasingly been leveraged to tackle agricultural optimization problems, and in particular, reinforcement learning (RL), a subfield of machine learning, seems a promising tool for data-driven discovery of future farm management policies. In this work, we present the development of CropGym, a Gymnasium environment, where a reinforcement learning agent can learn crop management policies using a variety of process-based crop growth models. As a use case, we report on the discovery of strategies for nitrogen application in winter wheat. An RL agent is trained to decide weekly on applying a discrete amount of nitrogen fertilizer, with the aim of achieving a balance between maximizing yield and minimizing environmental impact. Results show that close to optimal strategies are learned, competitive with standard practices set by domain experts. In addition, we evaluate, as an out-of-distribution test, whether the obtained policies are resilient against a change in climate conditions. We find that, when rainfall is sufficient, the RL agent remains close to the optimal policy. With CropGym, we aim to facilitate collaboration between the RL and agronomy communities to address the challenges of future agricultural decision-making.
- 25 September 2023
- Published by Cambridge University Press (CUP) in Environmental Data Science
- Vol. 2, e34
Open AccessNitrogen management with reinforcement learning and crop growth models
The growing need for agricultural products and the challenges posed by environmental and economic factors have created a demand for enhanced agricultural systems management. Machine learning has increasingly been leveraged to tackle agricultural optimization problems, and in particular, reinforcement learning (RL), a subfield of machine learning, seems a promising tool for data-driven discovery of future farm management policies. In this work, we present the development of CropGym, a Gymnasium environment, where a reinforcement learning agent can learn crop management policies using a variety of process-based crop growth models. As a use case, we report on the discovery of strategies for nitrogen application in winter wheat. An RL agent is trained to decide weekly on applying a discrete amount of nitrogen fertilizer, with the aim of achieving a balance between maximizing yield and minimizing environmental impact. Results show that close to optimal strategies are learned, competitive with standard practices set by domain experts. In addition, we evaluate, as an out-of-distribution test, whether the obtained policies are resilient against a change in climate conditions. We find that, when rainfall is sufficient, the RL agent remains close to the optimal policy. With CropGym, we aim to facilitate collaboration between the RL and agronomy communities to address the challenges of future agricultural decision-making.
- Open Access
Integrating processed-based models and machine learning for crop yield prediction
Crop yield prediction typically involves the utilization of either theory-driven process-based crop growth models, which have proven to be difficult to calibrate for local conditions, or data-driven machine learning methods, which are known to require large datasets. In this work we investigate potato yield prediction using a hybrid meta-modeling approach. A crop growth model is employed to generate synthetic data for (pre)training a convolutional neural net, which is then fine-tuned with observational data. When applied in silico, our meta-modeling approach yields better predictions than a baseline comprising a purely data-driven approach. When tested on real-world data from field trials (n=303) and commercial fields (n=77), the meta-modeling approach yields competitive results with respect to the crop growth model. In the latter set, however, both models perform worse than a simple linear regression with a hand-picked feature set and dedicated preprocessing designed by domain experts. Our findings indicate the potential of meta-modeling for accurate crop yield prediction; however, further advancements and validation using extensive real-world datasets is recommended to solidify its practical effectiveness.
Open AccessIntegrating processed-based models and machine learning for crop yield prediction
Crop yield prediction typically involves the utilization of either theory-driven process-based crop growth models, which have proven to be difficult to calibrate for local conditions, or data-driven machine learning methods, which are known to require large datasets. In this work we investigate potato yield prediction using a hybrid meta-modeling approach. A crop growth model is employed to generate synthetic data for (pre)training a convolutional neural net, which is then fine-tuned with observational data. When applied in silico, our meta-modeling approach yields better predictions than a baseline comprising a purely data-driven approach. When tested on real-world data from field trials (n=303) and commercial fields (n=77), the meta-modeling approach yields competitive results with respect to the crop growth model. In the latter set, however, both models perform worse than a simple linear regression with a hand-picked feature set and dedicated preprocessing designed by domain experts. Our findings indicate the potential of meta-modeling for accurate crop yield prediction; however, further advancements and validation using extensive real-world datasets is recommended to solidify its practical effectiveness.