Quantifying and Leveraging Classification Uncertainty for Chest Radiograph Assessment
Ghesu FC, Georgescu B, Gibson E, Gündel S, Kalra M, Singh R, Digumarthy S, Grbic S, Comaniciu D (2019)
Publication Type: Conference contribution, Conference Contribution
Publication year: 2019
Publisher: Springer International Publishing
Pages Range: 676-684
Conference Proceedings Title: Medical Image Computing and Computer Assisted Intervention – MICCAI 2019
Event location: Shenzhen, China
ISBN: 978-3-030-32226-7
DOI: 10.1007/978-3-030-32226-7_75
Abstract
The interpretation of chest radiographs is an essential task for the detection of thoracic diseases and abnormalities. However, it is a challenging problem with high inter-rater variability and inherent ambiguity due to inconclusive evidence in the data, limited data quality or subjective definitions of disease appearance. Current deep learning solutions for chest radiograph abnormality classification are typically limited to providing probabilistic predictions, relying on the capacity of learning models to adapt to the high degree of label noise and become robust to the enumerated causal factors. In practice, however, this leads to overconfident systems with poor generalization on unseen data. To account for this, we propose an automatic system that learns not only the probabilistic estimate on the presence of an abnormality, but also an explicit uncertainty measure which captures the confidence of the system in the predicted output. We argue that explicitly learning the classification uncertainty as an orthogonal measure to the predicted output, is essential to account for the inherent variability characteristic of this data. Experiments were conducted on two datasets of chest radiographs of over 85,000 patients. Sample rejection based on the predicted uncertainty can significantly improve the ROC-AUC, e.g., by 8% to 0.91 with an expected rejection rate of under 25%. Eliminating training samples using uncertainty-driven bootstrapping, enables a significant increase in robustness and accuracy. In addition, we present a multi-reader study showing that the predictive uncertainty is indicative of reader errors.
Authors with CRIS profile
Florin-Cristian Ghesu
Lehrstuhl für Informatik 14 (Bild- und Sprachverarbeitung)
Sebastian Gündel
Lehrstuhl für Informatik 5 (Mustererkennung)
Involved external institutions
Germany (DE)How to cite
APA:
Ghesu, F.-C., Georgescu, B., Gibson, E., Gündel, S., Kalra, M., Singh, R.,... Comaniciu, D. (2019). Quantifying and Leveraging Classification Uncertainty for Chest Radiograph Assessment. In Dinggang Shen,Tianming Liu,Terry M. Peters, Lawrence H. Staib, Caroline Essert, Sean Zhou, Pew-Thian Yap, Ali Khan (Eds.), Medical Image Computing and Computer Assisted Intervention – MICCAI 2019 (pp. 676-684). Shenzhen, China: Springer International Publishing.
MLA:
Ghesu, Florin-Cristian, et al. "Quantifying and Leveraging Classification Uncertainty for Chest Radiograph Assessment." Proceedings of the MICCAI: International Conference on Medical Image Computing and Computer-Assisted Intervention, Shenzhen, China Ed. Dinggang Shen,Tianming Liu,Terry M. Peters, Lawrence H. Staib, Caroline Essert, Sean Zhou, Pew-Thian Yap, Ali Khan, Springer International Publishing, 2019. 676-684.
BibTeX: Download