Overcoming the lack of annotated medical images by non‐expert annotation?
Clinical Neuroimaging · 2024
DOI: 10.1002/neo2.15 ↗Kurz gefasst
Fachartikel in der Zeitschrift Clinical Neuroimaging, 2024. Verfasst von Carsten M. Klingner und 6 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Deutsche Forschungsgemeinschaft.
Abstract (Originalsprache)
Objective
Early diagnosis of ischemic stroke is crucial. While CT scans are recommended, interpreting them is challenging. Machine learning can aid interpretation, but the lack of large datasets hinders its development. We explored if non‐expert labeling enhances the performance of a deep learning system trained on expert‐labeled datasets.
Methods
One‐hundred eight CT datasets were labeled for suspected strokes by both an expert and a non‐expert. A DenseNet121 model was trained on either expert data alone or with added non‐expert data. Both models were tested using expert‐only data, evaluating performance metrics such as accuracy, sensitivity, and F1 score.
Results
Expert and non‐expert labeling showed strong agreement (Cohen's kappa: 0.83). Adding non‐expert data enhanced model accuracy from 86.2% to 87.7% ( p < .001) and the F1 score from 86.5% to 88.4% for non‐stroke images, and from 85.7% to 87.0% for stroke images.
Conclusion
Machine learning performance in stroke detection can benefit from non‐expert labeled datasets. The implications for other medical conditions and the effect on larger datasets are yet to be determined. This could potentially address the issue of scarce labeled stroke imaging datasets.
Förderung
- Deutsche Forschungsgemeinschaft
Zitieren
APA
Klingner, C. M., Hemm, S., Wagner, F., Mayer, T. E., Guellmar, D., Witte, O. W., & Brodoehl, S. (2024). Overcoming the lack of annotated medical images by non‐expert annotation?. Clinical Neuroimaging. https://doi.org/10.1002/neo2.15
Vancouver
Klingner CM, Hemm S, Wagner F, Mayer TE, Guellmar D, Witte OW, et al. Overcoming the lack of annotated medical images by non‐expert annotation?. Clinical Neuroimaging. 2024. doi:10.1002/neo2.15
BibTeX
@article{Klingner2024,
author = {Klingner, Carsten M. and Hemm, Sophia and Wagner, Franziska and Mayer, Thomas E. and Guellmar, Daniel and Witte, Otto W. and Brodoehl, Stefan},
title = {Overcoming the lack of annotated medical images by non‐expert annotation?},
journal = {Clinical Neuroimaging},
year = {2024},
doi = {10.1002/neo2.15}
}
EUVIMED