Danyal Fareed Bhutto
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Danyal Fareed Bhutto has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Matthew Scott Rosen
A.A. Martinos Center for Biomedical Imaging, Department of Radiology (J.E.I., M.S.R.), Mass General Brigham and Harvard Medical School, Boston
2 shared publications - Aapo Nummenmaa
Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology Massachusetts General Hospital Boston Massachusetts USA
1 shared publication - Bo Zhu
Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Charlestown, MA, USA
1 shared publication - Bruce R. Rosen
Massachusetts General Hospital
1 shared publication - Declan M. Daly
University of Maryland, College Park
1 shared publication Dylan Edwards
Jefferson Moss Rehabilitation Research Institute Philadelphia Pennsylvania USA
1 shared publication
Latest Publications
- Open Access
- 30 May 2026
- Published by Wiley in Human Brain Mapping
- Vol. 47 (8) , e70550
Standardizing TMS Intensity Across Different Coils Using Individualized Electric Field Modeling
Quantitative measures for Transcranial Magnetic Stimulation (TMS) intensity are needed to ensure safe and consistent application in therapeutic and research settings. However, resting motor thresholds (rMTs), commonly used to determine stimulation intensity, depend on the coil used. Unless motor mapping and treatment coils are identical, re‐thresholding is necessary, increasing patient discomfort and potentially introducing variability across studies. These considerations raise an unresolved fundamental question: does individual rMT reflect a consistent cortical electric field (E‐field) magnitude independent of coil geometry? We tested the hypothesis that rMT corresponds to a coil‐invariant cortical E‐field magnitude and evaluated a computational method for predicting stimulator output across different coils using a reference rMT. Thirteen healthy, right‐handed participants were recruited; ten were included in the primary analysis. Participants underwent TMS with two figure‐of‐eight coils of different sizes. E‐field distributions were simulated using a fast multipole boundary element method, in free space and within personalized MRI‐based head models. rMT prediction accuracy was compared between a detailed five‐layer and a simplified three‐layer head model, and both were evaluated against direct rMT scaling using the reference coil. The personalized E‐field‐based approach significantly improved rMT prediction accuracy over direct scaling ( p < 0.001). The root‐mean‐square error (RMSE) was 1.26% and 1.32% of maximum stimulator output (MSO) for detailed and simplified models, versus 6.1% MSO for direct scaling. Individual rMT corresponds to a constant cortical E‐field magnitude ratio across coil types. E‐field‐based prediction offers a more accurate, coil‐independent method for standardizing TMS intensity, reducing the need for repeated thresholding.
- 30 May 2026
- Published by Wiley in Human Brain Mapping
- Vol. 47 (8) , e70550
Open AccessStandardizing TMS Intensity Across Different Coils Using Individualized Electric Field Modeling
Quantitative measures for Transcranial Magnetic Stimulation (TMS) intensity are needed to ensure safe and consistent application in therapeutic and research settings. However, resting motor thresholds (rMTs), commonly used to determine stimulation intensity, depend on the coil used. Unless motor mapping and treatment coils are identical, re‐thresholding is necessary, increasing patient discomfort and potentially introducing variability across studies. These considerations raise an unresolved fundamental question: does individual rMT reflect a consistent cortical electric field (E‐field) magnitude independent of coil geometry? We tested the hypothesis that rMT corresponds to a coil‐invariant cortical E‐field magnitude and evaluated a computational method for predicting stimulator output across different coils using a reference rMT. Thirteen healthy, right‐handed participants were recruited; ten were included in the primary analysis. Participants underwent TMS with two figure‐of‐eight coils of different sizes. E‐field distributions were simulated using a fast multipole boundary element method, in free space and within personalized MRI‐based head models. rMT prediction accuracy was compared between a detailed five‐layer and a simplified three‐layer head model, and both were evaluated against direct rMT scaling using the reference coil. The personalized E‐field‐based approach significantly improved rMT prediction accuracy over direct scaling ( p < 0.001). The root‐mean‐square error (RMSE) was 1.26% and 1.32% of maximum stimulator output (MSO) for detailed and simplified models, versus 6.1% MSO for direct scaling. Individual rMT corresponds to a constant cortical E‐field magnitude ratio across coil types. E‐field‐based prediction offers a more accurate, coil‐independent method for standardizing TMS intensity, reducing the need for repeated thresholding.
- Open Access
- 17 March 2025
- Published by American Physical Society (APS) in Physical Review Applied
- Vol. 23 (3)
Machine learning for improved current-density reconstruction from two-dimensional vector magnetic images
The reconstruction of electrical current densities from magnetic field measurements is an important technique with applications in materials science, circuit design, quality control, plasma physics, and biology. Analytic reconstruction methods exist for planar currents, but break down in the presence of high-spatial-frequency noise or large standoff distance, restricting the types of systems that can be studied. Here, we demonstrate the use of a deep convolutional neural network for current density reconstruction from two-dimensional images of vector magnetic fields acquired by a quantum diamond microscope . Trained network performance significantly exceeds analytic reconstruction for data with high noise or large standoff distances. This machine learning technique can perform quality inversions on lower-signal-to-noise-ratio data, significantly reducing the data collection time and permitting reconstructions of weaker and three-dimensional current sources.
- 17 March 2025
- Published by American Physical Society (APS) in Physical Review Applied
- Vol. 23 (3)
Open AccessMachine learning for improved current-density reconstruction from two-dimensional vector magnetic images
The reconstruction of electrical current densities from magnetic field measurements is an important technique with applications in materials science, circuit design, quality control, plasma physics, and biology. Analytic reconstruction methods exist for planar currents, but break down in the presence of high-spatial-frequency noise or large standoff distance, restricting the types of systems that can be studied. Here, we demonstrate the use of a deep convolutional neural network for current density reconstruction from two-dimensional images of vector magnetic fields acquired by a quantum diamond microscope . Trained network performance significantly exceeds analytic reconstruction for data with high noise or large standoff distances. This machine learning technique can perform quality inversions on lower-signal-to-noise-ratio data, significantly reducing the data collection time and permitting reconstructions of weaker and three-dimensional current sources.
- 1 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 28 (9) , 5422-5434
Uncertainty Estimation and Out-of-Distribution Detection for Deep Learning-Based Image Reconstruction Using the Local Lipschitz
Accurate image reconstruction is at the heart of diagnostics in medical imaging. Supervised deep learning-based approaches have been investigated for solving inverse problems including image reconstruction. However, these trained models encounter unseen data distributions that are widely shifted from training data during deployment. Therefore, it is essential to assess whether a given input falls within the training data distribution. Current uncertainty estimation approaches focus on providing an uncertainty map to radiologists, rather than assessing the training distribution fit. In this work, we propose a method based on the local Lipschitz metric to distinguish out-of-distribution images from in-distribution with an area under the curve of 99.94% for True Positive Rate versus False Positive Rate. We demonstrate a very strong relationship between the local Lipschitz value and mean absolute error (MAE), supported by a Spearman's rank correlation coefficient of 0.8475, to determine an uncertainty estimation threshold for optimal performance. Through the identification of false positives, we demonstrate the local Lipschitz and MAE relationship can guide data augmentation and reduce uncertainty. Our study was validated using the AUTOMAP architecture for sensor-to-image Magnetic Resonance Imaging (MRI) reconstruction. We demonstrate our approach outperforms baseline techniques of Monte-Carlo dropout and deep ensembles as well as the state-of-the-art Mean Variance Estimation (MVE) network approach. We expand our application scope to MRI denoising and Computed Tomography (CT) sparse-to-full view reconstructions using UNET architectures. We show our approach is applicable to various architectures and applications, especially in medical imaging, where preserving diagnostic accuracy of reconstructed images remains paramount.
- 1 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 28 (9) , 5422-5434
Uncertainty Estimation and Out-of-Distribution Detection for Deep Learning-Based Image Reconstruction Using the Local Lipschitz
Accurate image reconstruction is at the heart of diagnostics in medical imaging. Supervised deep learning-based approaches have been investigated for solving inverse problems including image reconstruction. However, these trained models encounter unseen data distributions that are widely shifted from training data during deployment. Therefore, it is essential to assess whether a given input falls within the training data distribution. Current uncertainty estimation approaches focus on providing an uncertainty map to radiologists, rather than assessing the training distribution fit. In this work, we propose a method based on the local Lipschitz metric to distinguish out-of-distribution images from in-distribution with an area under the curve of 99.94% for True Positive Rate versus False Positive Rate. We demonstrate a very strong relationship between the local Lipschitz value and mean absolute error (MAE), supported by a Spearman's rank correlation coefficient of 0.8475, to determine an uncertainty estimation threshold for optimal performance. Through the identification of false positives, we demonstrate the local Lipschitz and MAE relationship can guide data augmentation and reduce uncertainty. Our study was validated using the AUTOMAP architecture for sensor-to-image Magnetic Resonance Imaging (MRI) reconstruction. We demonstrate our approach outperforms baseline techniques of Monte-Carlo dropout and deep ensembles as well as the state-of-the-art Mean Variance Estimation (MVE) network approach. We expand our application scope to MRI denoising and Computed Tomography (CT) sparse-to-full view reconstructions using UNET architectures. We show our approach is applicable to various architectures and applications, especially in medical imaging, where preserving diagnostic accuracy of reconstructed images remains paramount.