Deep Neural Networks (DNNs) can provide clinicians with fast and accurate predictions that are highly valuable for high-stakes medical decision-making, such as in brain tumor segmentation and treatment planning. However, these models largely lack transparency about the uncertainty in their predictions, potentially giving clinicians a false sense of reliability that may lead to grave consequences in patient care. Growing calls for Transparent and Responsible AI have promoted Uncertainty Quantification (UQ) to capture and communicate uncertainty in a systematic and principled manner. However, traditional Bayesian UQ methods remain prohibitively costly for large, million-dimensional tumor segmentation DNNs such as the U-Net. In this work, we discuss a computationally-efficient UQ approach via the partially Bayesian neural networks (pBNN). In pBNN, only a single layer, strategically selected based on gradient-based sensitivity analysis, is targeted for Bayesian inference. We illustrate the effectiveness of pBNN in capturing the full uncertainty for a 7.8-million parameter U-Net. We also demonstrate how practitioners and model developers can use the pBNN's predictions to better understand the model's capabilities and behavior.
Prabhudesai, Snehal, et al. "Lowering the computational barrier: Partially Bayesian neural networks for transparency in medical imaging AI." Frontiers in Computer Science, vol. 5, Feb. 2023. https://doi.org/10.3389/fcomp.2023.1071174
Prabhudesai, Snehal, Hauth, Jeremiah, Guo, Dingkun, et al., "Lowering the computational barrier: Partially Bayesian neural networks for transparency in medical imaging AI," Frontiers in Computer Science 5 (2023), https://doi.org/10.3389/fcomp.2023.1071174
@article{osti_1924814,
author = {Prabhudesai, Snehal and Hauth, Jeremiah and Guo, Dingkun and Rao, Arvind and Banovic, Nikola and Huan, Xun},
title = {Lowering the computational barrier: Partially Bayesian neural networks for transparency in medical imaging AI},
annote = {Deep Neural Networks (DNNs) can provide clinicians with fast and accurate predictions that are highly valuable for high-stakes medical decision-making, such as in brain tumor segmentation and treatment planning. However, these models largely lack transparency about the uncertainty in their predictions, potentially giving clinicians a false sense of reliability that may lead to grave consequences in patient care. Growing calls for Transparent and Responsible AI have promoted Uncertainty Quantification (UQ) to capture and communicate uncertainty in a systematic and principled manner. However, traditional Bayesian UQ methods remain prohibitively costly for large, million-dimensional tumor segmentation DNNs such as the U-Net. In this work, we discuss a computationally-efficient UQ approach via the partially Bayesian neural networks (pBNN). In pBNN, only a single layer, strategically selected based on gradient-based sensitivity analysis, is targeted for Bayesian inference. We illustrate the effectiveness of pBNN in capturing the full uncertainty for a 7.8-million parameter U-Net. We also demonstrate how practitioners and model developers can use the pBNN's predictions to better understand the model's capabilities and behavior.},
doi = {10.3389/fcomp.2023.1071174},
url = {https://www.osti.gov/biblio/1924814},
journal = {Frontiers in Computer Science},
issn = {ISSN 2624-9898},
volume = {5},
place = {Switzerland},
publisher = {Frontiers Media SA},
year = {2023},
month = {02}}
Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part IIIhttps://doi.org/10.1007/978-3-319-24574-4_28
FAT* '19: Conference on Fairness, Accountability, and Transparency, Proceedings of the Conference on Fairness, Accountability, and Transparencyhttps://doi.org/10.1145/3287560.3287596