Mitigating Bias in Radiology Machine Learning: 3. Performance Metrics
Open accessThe DOI matches its Crossref record; 22 of the 22 references it cites were found in a registry or an index.
Abstract
The increasing use of machine learning (ML) algorithms in clinical settings raises concerns about bias in ML models. Bias can arise at any step of ML creation, including data handling, model development, and performance evaluation. Potential biases in the ML model can be minimized by implementing these steps correctly. This report focuses on performance evaluation and discusses model fitness, as well as a set of performance evaluation toolboxes: namely, performance metrics, performance interpretation maps, and uncertainty quantification. By discussing the strengths and limitations of each toolbox, our report highlights strategies and considerations to mitigate and detect biases during performance evaluations of radiology artificial intelligence models. Keywords: Segmentation, Diagnosis, Convolutional Neural Network (CNN) © RSNA, 2022
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Rudin, C. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0048-x · Cited by 10,561 · cited 1×
“…post hoc attri-bution, trainable attention modules can be attached to typical convolutional neural network architectures to produce so-called attention maps during model training (50). interpretation methods, such as saliency mapping or class activation maps, are unreliable and misleading (52,53).”
BackgroundDissecting racial bias in an algorithm used to manage the health of populations
Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019) · Science · doi:10.1126/science.aax2342 · Cited by 7,137
A review of uncertainty quantification in deep learning: Techniques, applications and challenges
Abdar, M., Pourpanah, F., Hussain, S., Rezazadegan, D., Liu, L., Ghavamzadeh, M., Fieguth, P., Cao, X., Khosravi, A., Acharya, U. R., et al. (2021) · Information Fusion · doi:10.1016/j.inffus.2021.05.008 · Cited by 3,111 · cited 1×
“Although there are many UQ techniques in ML research, the more popular techniques may be classified into three groups (64,65): 1.”
MethodAleatory or epistemic? Does it matter?
Kiureghian, A. D., & Ditlevsen, O. (2009) · Structural Safety · doi:10.1016/j.strusafe.2008.06.020 · Cited by 2,454 · cited 1×
“Usually, model uncertainty refers to epistemic uncertainty, which is focused on here (59,60).”
BackgroundImplementing Machine Learning in Health Care — Addressing Ethical Challenges
Char, D. S., Shah, N. H., & Magnus, D. (2018) · New England Journal of Medicine · doi:10.1056/nejmp1714229 · Cited by 1,857 · cited 5×
“As a result, the generative task described by Equation (5) is illposed (46) and often reformulated as an optimization problem:”
BackgroundInfluentialDo no harm: a roadmap for responsible machine learning for health care
Wiens, J., Saria, S., Sendak, M., Ghassemi, M., Liu, V. X., Doshi-Velez, F., Jung, K., Heller, K., Kale, D., Saeed, M., et al. (2019) · Nature Medicine · doi:10.1038/s41591-019-0548-6 · Cited by 1,266
Trust in a specific technology: An investigation of its components and measures
Mcknight, D. H., Carter, M., Thatcher, J. B., & Clay, P. F. (2011) · ACM Transactions on Management Information Systems · doi:10.1145/1985347.1985353 · Cited by 1,185
Clinical decision support in the era of artificial intelligence
Shortliffe, E. H., & Sepúlveda, M. J. (2018) · JAMA · doi:10.1001/jama.2018.17163 · Cited by 920 · cited 2×
“10 (max ) PSNR 1 0 log ( ) MSE d = (8),”
BackgroundPark, S. H., & Han, K. (2018) · Radiology · doi:10.1148/radiol.2017171920 · Cited by 894 · cited 1×
“Although these techniques can help determine model fitness, internal testing may not detect biases present in the original data (19,20).”
BackgroundLarrazabal, A. J., Nieto, N., Peterson, V., Milone, D. H., & Ferrante, E. (2020) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1919012117 · Cited by 685 · cited 1×
“Larrazabal et al (22) showed that the performance of classifier models trained on the National Institutes of Health ChestX-Ray14 dataset differs significantly if applied to chest radiograph data from other sources.”
MethodThe need for uncertainty quantification in machine-assisted medical decision making
Begoli, E., Bhattacharya, T., & Kusnezov, D. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-018-0004-1 · Cited by 597 · cited 1×
“In this way, the model communicates with medical experts and likely increases trust in model performance over time (57,58).”
BackgroundSecond opinion needed: Communicating uncertainty in medical machine learning
Kompa, B., Snoek, J., & Beam, A. L. (2021) · npj Digital Medicine · doi:10.1038/s41746-020-00367-3 · Cited by 514 · cited 1×
“In this way, the model communicates with medical experts and likely increases trust in model performance over time (57,58).”
BackgroundOzenne, B., Subtil, F., & Maucort-Boulch, D. (2015) · Journal of Clinical Epidemiology · doi:10.1016/j.jclinepi.2015.02.010 · Cited by 379
Assessing the trustworthiness of saliency maps for localizing abnormalities in medical imaging
Arun, N., Gaw, N., Singh, P., Chang, K., Aggarwal, M., Chen, B., Hoebel, K., Gupta, S., Patel, J., Gidwani, M., et al. (2021) · Radiology: Artificial Intelligence · doi:10.1148/ryai.2021200267 · Cited by 342 · cited 1×
“In a recent study, Arun et al (56) assessed interpretation maps using four critical trustworthiness criteria: utility, sensitivity to weight randomization, repeatability (intra-architecture), and reproducibility (interarchitecture).”
BackgroundBöhle, M., Eitel, F., Weygandt, M., & Ritter, K. (2019) · Frontiers in Aging Neuroscience · doi:10.3389/fnagi.2019.00194 · Cited by 291 · cited 1×
“Additionally, Böhle et al (55) found that guided backpropagation did not produce interpretation maps that were visually distinguishable in Alzheimer disease versus healthy controls using T1-weighted brain MRI studies.”
BackgroundDeep learning predicts hip fracture using confounding patient and healthcare variables
Badgeley, M. A., Zech, J. R., Oakden-Rayner, L., Glicksberg, B. S., Liu, M., Gale, W., McConnell, M. V., Percha, B., Snyder, T. M., & Dudley, J. T. (2019) · npj Digital Medicine · doi:10.1038/s41746-019-0105-1 · Cited by 290
Ho, S. Y., Phua, K., Wong, L., & Bin Goh, W. W. (2020) · Patterns · doi:10.1016/j.patter.2020.100129 · Cited by 247 · cited 1×
“Although these techniques can help determine model fitness, internal testing may not detect biases present in the original data (19,20).”
BackgroundTreating health disparities with artificial intelligence
Chen, I. Y., Joshi, S., & Ghassemi, M. (2020) · Nature Medicine · doi:10.1038/s41591-019-0649-2 · Cited by 235
Interpretation and visualization techniques for deep learning models in medical imaging
Huff, D. T., Weisman, A. J., & Jeraj, R. (2021) · Physics in Medicine & Biology · doi:10.1088/1361-6560/abcd17 · Cited by 172
Eche, T., Schwartz, L. H., Mokrane, F.-Z., & Dercle, L. (2021) · Radiology: Artificial Intelligence · doi:10.1148/ryai.2021210097 · Cited by 163
Mitigating Bias in Radiology Machine Learning: 1. Data Handling
Rouzrokh, P., Khosravi, B., Faghani, S., Moassefi, M., Vera Garcia, D. V., Singh, Y., Zhang, K., Conte, G. M., & Erickson, B. J. (2022) · Radiology: Artificial Intelligence · doi:10.1148/ryai.210290 · Cited by 145 · cited 1×
“There are different techniques to internally test an ML model, for example, simple train-validation-test split, cross-validation, and nested cross-validation (15).”
MethodSelf-citationTrust in artificial intelligence for medical diagnoses
(2020) · Progress in Brain Research · doi:10.1016/bs.pbr.2020.06.006 · Cited by 116
Cited by
AI pitfalls and what not to do: mitigating bias in AI
Gichoya, J. W., Thomas, K., Celi, L. A., Safdar, N., Banerjee, I., Banja, J. D., Seyyed-Kalantari, L., Trivedi, H., & Purkayastha, S. (2023) · The British Journal of Radiology · doi:10.1259/bjr.20230023 · Cited by 277
Drukker, K., Chen, W., Gichoya, J. W., Gruszauskas, N. P., Kalpathy-Cramer, J., Koyejo, S., Myers, K. J., Sá, R. C., Sahiner, B., Whitney, H. M., et al. (2023) · Journal of Medical Imaging · doi:10.1117/1.jmi.10.6.061104 · Cited by 149
Understanding and Mitigating Bias in Imaging Artificial Intelligence
Tejani, A. S., Ng, Y. S., Xi, Y., & Rayan, J. C. (2024) · RadioGraphics · doi:10.1148/rg.230067 · Cited by 138
Hanneman, K., Playford, D., Dey, D., van Assen, M., Mastrodicasa, D., Cook, T. S., Gichoya, J. W., Williamson, E. E., Rubin, G. D., & on behalf of the American Heart Association Council on Cardiovascular Radiology and Intervention; and Council on Lifelong Congenital Heart Disease and Heart Health in the Young (2024) · Circulation · doi:10.1161/cir.0000000000001202 · Cited by 72
“71”
Mastrodicasa, D., van Assen, M., Huisman, M., Leiner, T., Williamson, E. E., Nicol, E. D., Allen, B. D., Saba, L., Vliegenthart, R., & Hanneman, K. (2025) · Radiology · doi:10.1148/radiol.240516 · Cited by 69
Metadata
- Authors
- Faghani, S., Khosravi, B., Zhang, K., Moassefi, M., Jagtap, J. M., Nugen, F., Vahdati, S., Kuanar, S. P., Rassoulinejad-Mousavi, S. M., Singh, Y., Vera Garcia, D. V., Rouzrokh, P., & Erickson, B. J.
- Venue
- Radiology: Artificial Intelligence
- Published
- 2022-09-01
- Volume
- 4
- Issue
- 5
- Pages
- e220061-e220061
- ISSN
- 2638-6100 (ISSN-L)
- OpenAlex
- W4293004949
- Cited by
- 94
- References
- 22
- FWCI
- 4.37
- Type
- journal article
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.