Why Are Explainable AI Methods for Prostate Lesion Detection Rated Poorly by Radiologists?
Source abstract
Abstract
Deep learning offers significant advancements in the accuracy of prostate identification and classification, underscoring its potential for clinical integration. However, the opacity of deep learning models presents interpretability challenges, critical for their acceptance and utility in medical diagnosis and detection. While explanation methods have been proposed to demystify these models, enhancing their clinical viability, the efficacy and acceptance of these methods in medical tasks are not well documented. This pilot study investigates the effectiveness of deep learning explanation methods in clinical settings and identifies the attributes that radiologists consider crucial for explainability, aiming to direct future enhancements. This study reveals that while explanation methods can improve clinical task performance by up to 20%, their perceived usefulness varies, with some methods being rated poorly. Radiologists prefer explanation methods that are robust against noise, precise, and consistent. These preferences underscore the need for refining explanation methods to align with clinical expectations, emphasizing clarity, accuracy, and reliability. The findings highlight the importance of developing explanation methods that not only improve performance but also are tailored to meet the stringent requirements of clinical practice, thereby facilitating deeper trust and a broader acceptance of deep learning in medical diagnostics.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Artificial Intelligence in Healthcare and Education · Medical Imaging and Analysis · Radiomics and Machine Learning in Medical Imaging
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Computer Science & Engineering Department, J.B. Speed School of Engineering, University of Louisville, Louisville, KY 40292, USA
Department of Radiology, Division of Interventional Radiology, Mayo Clinic Arizona, Phoenix, AZ 85054, USA
Department of Radiology, Canakkale Onsekiz Mart University, 17100 Canakkale, Turkey
Department of Radiology, Gaziantep University, 27310 Gaziantep, Turkey
References (38)
Show source-provided references
1. A Deep Learning-Based Approach for the Detection and Localization of Prostate Cancer in T2 Magnetic Resonance Images10.1007/s10278-018-0160-1 ↗
2. Classification of Cancer at Prostate MRI: Deep Learning versus Clinical PI-RADS Assessment10.1148/radiol.2019190938 ↗
3. Deep Learning Regression for Prostate Cancer Detection and Grading in Bi-Parametric MRI10.1109/TBME.2020.2993528 ↗
4. Meaningful Information and the Right to Explanation10.1093/idpl/ipx022 ↗
5. Building Trust in Deep Learning System towards Automated Disease Detection
6. Salahuddin, Z., Woodruff, H., Chatterjee, A., and Lambin, P. (2022). Transparency of Deep Neural Networks for Medical Image Analysis: A Review of Interpretability Methods. Comput. Biol. Med., 140.10.1016/j.compbiomed.2021.105111 ↗
7. Ribeiro, M., Singh, S., and Guestrin, C. (2016, January 12–17). “Why Should I Trust You?”: Explaining the Predictions of Any Classifier. Proceedings of the NAACL 2016, San Diego, CA, USA.10.18653/v1/N16-3020 ↗
8. Binder, A., Montavon, G., Lapuschkin, S., Müller, K., and Samek, W. (2016). Layer-Wise Relevance Propagation for Neural Networks with Local Renormalization Layers, ICANN.10.1007/978-3-319-44781-0_8 ↗
9. Ribeiro, M., Singh, S., and Guestrin, C. (2018). Anchors: High-Precision Model-Agnostic Explanations, AAAI.10.1609/aaai.v32i1.11491 ↗
10. Saralajew, S., Holdijk, L., Rees, M., Asan, E., and Villmann, T. (2019, January 8–14). Classification-by-Components: Probabilistic Modeling of Reasoning over a Set of Components. Proceedings of the NeurIPS 2019, Vancouver, BC, Canada.
11. Li, O., Liu, H., Chen, C., and Rudin, C. (2018). Deep Learning for Case-Based Reasoning through Prototypes: A Neural Network That Explains Its Predictions, AAAI.10.1609/aaai.v32i1.11771 ↗
12. Tonekaboni, S., Joshi, S., McCradden, M.D., and Goldenberg, A. (2019, January 9–10). What Clinicians Want: Contextualizing Explainable Machine Learning for Clinical End Use. Proceedings of the MLHC 2019, Ann Arbor, MI, USA.
13. Gulum, M.A., Trombley, C.M., and Kantardzic, M. (2021). A Review of Explainable Deep Learning Cancer Detection Models in Medical Imaging. Appl. Sci., 11.10.3390/app11104573 ↗
14. Hegselmann, S., Volkert, T., Ohlenburg, H., Gottschalk, A., Dugas, M., and Ertmer, C. (2020, January 7–8). An Evaluation of the Doctor-Interpretability of Generalized Additive Models with Interactions. Proceedings of the MLHC 2020, Durham, NC, USA.
15. The Stakes of Uncertainty: Developing and Integrating Machine Learning in Clinical Care
16. Piloting electronic medical record-based early detection of inpatient deterioration in community hospitals10.1002/jhm.2652 ↗
17. Toward a better understanding of task demands, workload, and performance during physician-computer interactions10.1093/jamia/ocw016 ↗
18. Ghassemi, M., Pushkarna, M., Wexler, J., Johnson, J., and Varghese, P. (2018). ClinicalVis: Supporting Clinical Task-Focused Design Evaluation. arXiv.
19. Lertvittayakumjorn, P., and Toni, F. (2019). Human-grounded Evaluations of Explanation Methods for Text Classification. arXiv.10.18653/v1/D19-1523 ↗
20. Shrikumar, A., Greenside, P., and Kundaje, A. (2017, January 6–11). Learning important features through propagating activation differences. Proceedings of the 34th International Conference on Machine Learning—Volume 70 (ICML’17), Sydney, Australia.
21. Jeyakumar, J., Noor, J., Cheng, Y., Garcia, L., and Srivastava, M.B. (2020, January 6–12). How Can I Explain This to You? An Empirical Study of Deep Neural Network Explanation Methods. Proceedings of the NeurIPS 2020, Online.
22. Chen, C., Li, O., Barnett, A., Su, J., and Rudin, C. (2019, January 8–14). This looks like that: Deep learning for interpretable image recognition. Proceedings of the NeurIPS 2019, Vancouver, BC, Canada.
23. Krosnick, J.A., and Presser, S. (2009). Question and Questionnaire Design. Handbook of Survey Research, Stanford University. [2nd ed.].
24. PROSTATEx Challenge Data
25. Quality Control and Whole-Gland, Zonal and Lesion Annotations for the PROSTATEx Challenge Public Dataset10.1016/j.ejrad.2021.109647 ↗
26. Simonyan, K., and Zisserman, A. (2015). Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv.
27. Prostate Cancer Diagnosis using Deep Learning with 3D Multiparametric MRI
28. Searching for Prostate Cancer by Fully Automated Magnetic Resonance Imaging Classification: Deep Learning versus Non-Deep Learning
29. Classification of Clinical Significance of MRI Prostate Findings Using 3D Convolutional Neural Networks
30. Sundararajan, M., Taly, A., and Yan, Q. (2017, January 6–11). Axiomatic Attribution for Deep Networks. Proceedings of the 34th International Conference on Machine Learning—Volume 70, Sydney, NSW, Australia.
31. Smilkov, D., Thorat, N., Kim, B., Viégas, F.B., and Wattenberg, M. (2017). SmoothGrad: Removing Noise by Adding Noise. arXiv.
32. Lundberg, S.M., and Lee, S.I. (2017, January 4–9). A Unified Approach to Interpreting Model Classifications. Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS’17), Red Hook, NY, USA.
33. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization10.1007/s11263-019-01228-7 ↗
34. Schwab, P., and Karlen, W. (2019, January 8–14). CXPlain: Causal Explanations for Model Interpretation under Uncertainty. Proceedings of the Advances in Neural Information Processing Systems, Vancouver, BC, Canada.
35. Martel, A.L., Abolmaesumi, P., Stoyanov, D., Mateus, D., Zuluaga, M.A., Zhou, S.K., Racoceanu, D., and Joskowicz, L. (2020). Efficient Shapley Explanation for Features Importance Estimation Under Uncertainty. Medical Image Computing and Computer Assisted Intervention—MICCAI 2020, Springer International Publishing.10.1007/978-3-030-59728-3 ↗
36. Bykov, K., Höhne, M.M.-C., Müller, K.-R., Nakajima, S., and Kloft, M. (2020). How Much Can I Trust You?—Quantifying Uncertainties in Explaining Neural Networks. arXiv.
37. Zhao, X., Huang, W., Huang, X., Robu, V., and Flynn, D. (2021, January 27–30). BayLIME: Bayesian Local Interpretable Model-Agnostic Explanations. Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence, Proceedings of Machine Learning Research, Virtual Event.
38. A Pragmatic View of Thematic Analysis