Corpus ID: 211066570

On the limits of cross-domain generalization in automated X-ray prediction

@article{Cohen2020OnTL,
  title={On the limits of cross-domain generalization in automated X-ray prediction},
  author={Joseph Paul Cohen and Mohammad Hashir and Rupert Brooks and Hadrien Bertrand},
  journal={ArXiv},
  year={2020},
  volume={abs/2002.02497}
}
  • Joseph Paul Cohen, Mohammad Hashir, +1 author Hadrien Bertrand
  • Published 2020
  • Engineering, Computer Science, Biology, Mathematics
  • ArXiv
  • This large scale study focuses on quantifying what X-rays diagnostic prediction tasks generalize well across multiple different datasets. We present evidence that the issue of generalization is not due to a shift in the images but instead a shift in the labels. We study the cross-domain performance, agreement between models, and model representations. We find interesting discrepancies between performance and agreement where models which both achieve good performance disagree in their… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    Explore Further: Topics Discussed in This Paper

    A Critic Evaluation of Methods for COVID-19 Automatic Detection from X-Ray Images
    • 15
    • Highly Influenced
    • PDF
    COVID-19 Image Data Collection: Prospective Predictions Are the Future
    • 25
    • PDF
    On the Composition and Limitations of Publicly Available COVID-19 X-Ray Imaging Datasets
    • 1
    • PDF
    Deep Mining External Imperfect Data for Chest X-ray Disease Screening
    • 1
    • PDF
    Predicting COVID-19 Pneumonia Severity on Chest X-ray With Deep Learning
    • 22
    • PDF
    Potential Features of ICU Admission in X-ray Images of COVID-19 Patients
    • 1
    • PDF

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 46 REFERENCES
    Comparison of Deep Learning Approaches for Multi-Label Chest X-Ray Classification
    • 66
    • PDF
    Can we trust deep learning models diagnosis? The impact of domain shift in chest radiograph classification
    • 10
    • PDF
    Evaluation: from Precision, Recall and F-measure to ROC, Informedness, Markedness and Correlation
    • 3,481
    • PDF
    CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparison
    • 239
    • Highly Influential
    • PDF
    Measuring the tendency of CNNs to Learn Surface Statistical Regularities
    • 102
    • PDF
    A Strong Baseline for Domain Adaptation and Generalization in Medical Imaging
    • 15
    • PDF
    Exploring large scale public medical image datasets
    • 24
    • PDF
    Densely Connected Convolutional Networks
    • 9,713
    • PDF