Acute cardiac care & resuscitation
Automated detection of severe cerebral edema using explainable deep transfer learning after hypoxic ischemic brain injury
Authors' conclusion
Our results support the feasibility of automated SCE detection. Future prospective studies with standardized neurologic assessments are needed to substantiate the utility of quantitative ML_SCE values to improve neuroprognostication.
Study context
- Population
- Cardiac arrest patients who remain comatose after the restoration of spontaneous circulation
- Participants
- 288 patients
- Comparison
- Automated machine learning (ML) detection of severe cerebral edema (SCE) (ML_SCE) against ground-truth SCE (GT_SCE) classification based on radiology reports
- Outcome
- Detection of SCE (cross-validated testing accuracy)
Methodological detailsmachine learning (ML), cross-validated
- Machine learning (ML)
- Cross-validated
Main finding
As reported in the abstract
Cross-validated testing accuracy
87%
95% CI 84%, 89%
Further findings
Attention maps explaining SCE classification
- Focused on cisternal regions
- P < 0.05
Predictors of poor neurologic outcomes (multivariable analyses)
- Older age
- P < 0.001
- Non-shockable initial cardiac rhythm
- P = 0.004
- Greater ML_SCE values
- P < 0.001
- GT_SCE (non-significant covariate)
- P = 0.064
- Multivariable analyses
Original publication
Automated detection of severe cerebral edema using explainable deep transfer learning after hypoxic ischemic brain injury. Resuscitation. Published 23 May 2025. doi:10.
Summarised from the published abstract; every number is checked against it