RT Journal Article
T1 Improving Uncertainty Estimation With Semi-Supervised Deep Learning for COVID-19 Detection Using Chest X-Ray Images.
A1 Calderon-Ramirez, Saul
A1 Yang, Shengxiang
A1 Moemeni, Armaghan
A1 Colreavy-Donnelly, Simon
A1 Elizondo, David A
A1 Oala, Luis
A1 Rodriguez-Capitan, Jorge
A1 Jimenez-Navarro, Manuel
A1 Lopez-Rubio, Ezequiel
A1 Molina-Cabello, Miguel A
K1 Coronavirus
K1 Covid-19
K1 MixMatch
K1 Uncertainty estimation
K1 chest x-ray
K1 computer aided diagnosis
K1 semi-supervised deep learning
AB In this work we implement a COVID-19 infection detection system based on chest X-ray images with uncertainty estimation. Uncertainty estimation is vital for safe usage of computer aided diagnosis tools in medical applications. Model estimations with high uncertainty should be carefully analyzed by a trained radiologist. We aim to improve uncertainty estimations using unlabelled data through the MixMatch semi-supervised framework. We test popular uncertainty estimation approaches, comprising Softmax scores, Monte-Carlo dropout and deterministic uncertainty quantification. To compare the reliability of the uncertainty estimates, we propose the usage of the Jensen-Shannon distance between the uncertainty distributions of correct and incorrect estimations. This metric is statistically relevant, unlike most previously used metrics, which often ignore the distribution of the uncertainty estimations. Our test results show a significant improvement in uncertainty estimates when using unlabelled data. The best results are obtained with the use of the Monte Carlo dropout method.
SN 2169-3536
YR 2021
FD 2021-06-02
LK https://hdl.handle.net/10668/25277
UL https://hdl.handle.net/10668/25277
LA en
DS RISalud
RD Apr 8, 2025