Automated Assessment of Brain CT After Cardiac Arrest-An Observational Derivation/Validation Cohort Study

Analysis of Cerebral CT Based on Supervised Machine Learning as a Predictor of Outcome After Out-of-Hospital Cardiac Arrest

BACKGROUND AND OBJECTIVES

In light of limited intensive care capacities and a lack of accurate prognostic tools to advise caregivers and family members responsibly, this study aims to determine whether automated cerebral CT (CCT) analysis allows prognostication after out-of-hospital cardiac arrest.

METHODS

In this monocentric, retrospective cohort study, a supervised machine learning classifier based on an elastic net regularized logistic regression model for gray matter alterations on nonenhanced CCT obtained after cardiac arrest was trained using 10-fold cross-validation and tested on a hold-out sample (random split 75%/25%) for outcome prediction. Following the literature, a favorable outcome was defined as a cerebral performance category of 1-2 and a poor outcome of 3-5. The diagnostic accuracy was compared with established and guideline-recommended prognostic measures within the sample, that is, gray matter-white matter ratio (GWR), neuron-specific enolase (NSE), and neurofilament light chain (NfL) in serum.

RESULTS

Of 279 adult patients, 132 who underwent CCT within 14 days of cardiac arrest with good imaging quality were identified. Our approach discriminated between favorable and poor outcomes with an area under the curve (AUC) of 0.73 (95% CI 0.59-0.82). Thus, the prognostic power outperformed the GWR (AUC 0.66, 95% CI 0.56-0.76). The biomarkers NfL, measured at days 1 and 2, and NSE, measured at day 2, exceeded the reliability of the imaging markers derived from CT (AUC NfL day 1: 0.87, 95% CI 0.75-0.99; AUC NfL day 2: 0.90, 95% CI 0.79-1.00; AUC NSE day: 2 0.78, 95% CI 0.62-0.94).

DISCUSSION

Our data show that machine learning-assisted gray matter analysis of CCT images offers prognostic information after out-of-hospital cardiac arrest. Thus, CCT gray matter analysis could become a reliable and time-independent addition to the standard workup with serum biomarkers sampled at predefined time points. Prospective studies are warranted to replicate these findings.

Regional Brain Net Water Uptake in Computed Tomography after Cardiac Arrest - A Novel Biomarker for Neuroprognostication

BACKGROUND

Selective water uptake by neurons and glial cells and subsequent brain tissue oedema are key pathophysiological processes of hypoxic-ischemic encephalopathy (HIE) after cardiac arrest (CA). Although brain computed tomography (CT) is widely used to assess the severity of HIE, changes of brain radiodensity over time have not been investigated. These could be used to quantify regional brain net water uptake (NWU), a potential prognostic biomarker.

METHODS

We conducted an observational prognostic accuracy study including a derivation (single center cardiac arrest registry) and a validation (international multicenter TTM2 trial) cohort. Early (<6 h) and follow-up (>24 h) head CTs of CA patients were used to determine regional NWU for grey and white matter regions after co-registration with a brain atlas. Neurological outcome was dichotomized as good versus poor using the Cerebral Performance Category Scale (CPC) in the derivation cohort and Modified Rankin Scale (mRS) in the validation cohort.

RESULTS

We included 115 patients (81 derivation, 34 validation) with out-of-hospital (OHCA) and in-hospital cardiac arrest (IHCA). Regional brain water content remained unchanged in patients with good outcome. In patients with poor neurological outcome, we found considerable regional water uptake with the strongest effect in the basal ganglia. NWU >8% in the putamen and caudate nucleus predicted poor outcome with 100% specificity (95%-CI: 86-100%) and 43% (moderate) sensitivity (95%-CI: 31-56%).

CONCLUSION

This pilot study indicates that NWU derived from serial head CTs is a promising novel biomarker for outcome prediction after CA. NWU >8% in basal ganglia grey matter regions predicted poor outcome while absence of NWU indicated good outcome. NWU and follow-up CTs should be investigated in larger, prospective trials with standardized CT acquisition protocols.

Standardised and automated assessment of head computed tomography reliably predicts poor functional outcome after cardiac arrest: a prospective multicentre study

PURPOSE

Application of standardised and automated assessments of head computed tomography (CT) for neuroprognostication after out-of-hospital cardiac arrest.

METHODS

Prospective, international, multicentre, observational study within the Targeted Hypothermia versus Targeted Normothermia after out-of-hospital cardiac arrest (TTM2) trial. Routine CTs from adult unconscious patients obtained > 48 h ≤ 7 days post-arrest were assessed qualitatively and quantitatively by seven international raters blinded to clinical information using a pre-published protocol. Grey-white-matter ratio (GWR) was calculated from four (GWR-4) and eight (GWR-8) regions of interest manually placed at the basal ganglia level. Additionally, GWR was obtained using an automated atlas-based approach. Prognostic accuracies for prediction of poor functional outcome (modified Rankin Scale 4-6) for the qualitative assessment and for the pre-defined GWR cutoff < 1.10 were calculated.

RESULTS

140 unconscious patients were included; median age was 68 years (interquartile range [IQR] 59-76), 76% were male, and 75% had poor outcome. Standardised qualitative assessment and all GWR models predicted poor outcome with 100% specificity (95% confidence interval [CI] 90-100). Sensitivity in median was 37% for the standardised qualitative assessment, 39% for GWR-8, 30% for GWR-4 and 41% for automated GWR. GWR-8 was superior to GWR-4 regarding prognostic accuracies, intra- and interrater agreement. Overall prognostic accuracy for automated GWR (area under the curve [AUC] 0.84, 95% CI 0.77-0.91) did not significantly differ from manually obtained GWR.

CONCLUSION

Standardised qualitative and quantitative assessments of CT are reliable and feasible methods to predict poor functional outcome after cardiac arrest. Automated GWR has the potential to make CT quantification for neuroprognostication accessible to all centres treating cardiac arrest patients.

Good outcome prediction after out-of-hospital cardiac arrest: A prospective multicenter observational study in Korea (the KORHN-PRO registry)

AIM

To assess the ability of clinical examination, biomarkers, electrophysiology and brain imaging, individually or in combination to predict good neurological outcomes at 6 months after CA.

METHODS

This was a retrospective analysis of the Korean Hypothermia Network Prospective Registry 1.0, which included adult out-of-hospital cardiac arrest (OHCA) patients (≥18 years). Good outcome predictors were defined as both pupillary light reflex (PLR) and corneal reflex (CR) at admission, Glasgow Coma Scale Motor score (GCS-M) >3 at admission, neuron-specific enolase (NSE) <17 µg/L at 24-72 h, a median nerve somatosensory evoked potential (SSEP) N20/P25 amplitude >4 µV, continuous background without discharges on electroencephalogram (EEG), and absence of anoxic injury on brain CT and diffusion-weighted imaging (DWI).

RESULTS

A total of 1327 subjects were included in the final analysis, and their median age was 59 years; among them, 412 subjects had a good neurological outcome at 6 months. GCS-M >3 at admission had the highest specificity of 96.7% (95% CI 95.3-97.8), and normal brain DWI had the highest sensitivity of 96.3% (95% CI 92.9-98.4). When the two predictors were combined, the sensitivities tended to decrease (ranging from 2.7-81.1%), and the specificities tended to increase, ranging from81.3-100%. Through the explorative variation of the 2021 European Resuscitation Council (ERC) and the European Society of Intensive Care Medicine (ESICM) prognostication strategy algorithms, good outcomes were predicted, with a specificity of 83.2% and a sensitivity of 83.5% in patients by the algorithm.

CONCLUSIONS

Clinical examination, biomarker, electrophysiology, and brain imaging predicted good outcomes at 6 months after CA. When the two predictors were combined, the specificity further improved. With the 2021 ERC/ESICM guidelines, the number of indeterminate patients and the uncertainty of prognostication can be reduced by using a good outcome prediction algorithm.

Quantitative analysis of apparent diffusion coefficients to predict neurological prognosis in cardiac arrest survivors: an observational derivation and internal-external validation study

BACKGROUND

This study aimed to validate apparent diffusion coefficient (ADC) values and thresholds to predict poor neurological outcomes in out-of-hospital cardiac arrest (OHCA) survivors by quantitatively analysing the ADC values via brain magnetic resonance imaging (MRI).

METHODS

This observational study used prospectively collected data from two tertiary academic hospitals. The derivation cohort comprised 70% of the patients randomly selected from one hospital, whereas the internal validation cohort comprised the remaining 30%. The external validation cohort used the data from another hospital, and the MRI data were restricted to scans conducted at 3 T within 72-96 h after an OHCA experience. We analysed the percentage of brain volume below a specific ADC value at 50-step intervals ranging from 200 to 1200 × 10-6 mm2/s, identifying thresholds that differentiate between good and poor outcomes. Poor neurological outcomes were defined as cerebral performance categories 3-5, 6 months after experiencing an OHCA.

RESULTS

A total of 448 brain MRI scans were evaluated, including a derivation cohort (n = 224) and internal/external validation cohorts (n = 96/128, respectively). The proportion of brain volume with ADC values below 450, 500, 550, 600, and 650 × 10-6 mm2/s demonstrated good to excellent performance in predicting poor neurological outcomes in the derivation group (area under the curve [AUC] 0.89-0.91), and there were no statistically significant differences in performances among the derivation, internal validation, and external validation groups (all P > 0.5). Among these, the proportion of brain volume with an ADC below 600 × 10-6 mm2/s predicted a poor outcome with a 0% false-positive rate (FPR) and 76% (95% confidence interval [CI] 68-83) sensitivity at a threshold of > 13.2% in the derivation cohort. In both the internal and external validation cohorts, when using the same threshold, a specificity of 100% corresponded to sensitivities of 71% (95% CI 58-81) and 78% (95% CI 66-87), respectively.

CONCLUSIONS

In this validation study, by consistently restricting the MRI types and timing during quantitative analysis of ADC values in brain MRI, we observed high reproducibility and sensitivity at a 0% FPR. Prospective multicentre studies are necessary to validate these findings.

Outcome prediction of cardiac arrest with automatically computed gray-white matter ratio on computed tomography images

BACKGROUND

This study aimed to develop an automated method to measure the gray-white matter ratio (GWR) from brain computed tomography (CT) scans of patients with out-of-hospital cardiac arrest (OHCA) and assess its significance in predicting early-stage neurological outcomes.

METHODS

Patients with OHCA who underwent brain CT imaging within 12 h of return of spontaneous circulation were enrolled in this retrospective study. The primary outcome endpoint measure was a favorable neurological outcome, defined as cerebral performance category 1 or 2 at hospital discharge. We proposed an automated method comprising image registration, K-means segmentation, segmentation refinement, and GWR calculation to measure the GWR for each CT scan. The K-means segmentation and segmentation refinement was employed to refine the segmentations within regions of interest (ROIs), consequently enhancing GWR calculation accuracy through more precise segmentations.

RESULTS

Overall, 443 patients were divided into derivation N=265, 60% and validation N=178, 40% sets, based on age and sex. The ROI Hounsfield unit values derived from the automated method showed a strong correlation with those obtained from the manual method. Regarding outcome prediction, the automated method significantly outperformed the manual method in GWR calculation (AUC 0.79 vs. 0.70) across the entire dataset. The automated method also demonstrated superior performance across sensitivity, specificity, and positive and negative predictive values using the cutoff value determined from the derivation set. Moreover, GWR was an independent predictor of outcomes in logistic regression analysis. Incorporating the GWR with other clinical and resuscitation variables significantly enhanced the performance of prediction models compared to those without the GWR.

CONCLUSIONS

Automated measurement of the GWR from non-contrast brain CT images offers valuable insights for predicting neurological outcomes during the early post-cardiac arrest period.

Neuromonitoring in the ICU - what, how and why?

PURPOSE OF REVIEW

We selectively review emerging noninvasive neuromonitoring techniques and the evidence that supports their use in the ICU setting. The focus is on neuromonitoring research in patients with acute brain injury.

RECENT FINDINGS

Noninvasive intracranial pressure evaluation with optic nerve sheath diameter measurements, transcranial Doppler waveform analysis, or skull mechanical extensometer waveform recordings have potential safety and resource-intensity advantages when compared to standard invasive monitors, however each of these techniques has limitations. Quantitative electroencephalography can be applied for detection of cerebral ischemia and states of covert consciousness. Near-infrared spectroscopy may be leveraged for cerebral oxygenation and autoregulation computation. Automated quantitative pupillometry and heart rate variability analysis have been shown to have diagnostic and/or prognostic significance in selected subtypes of acute brain injury. Finally, artificial intelligence is likely to transform interpretation and deployment of neuromonitoring paradigms individually and when integrated in multimodal paradigms.

SUMMARY

The ability to detect brain dysfunction and injury in critically ill patients is being enriched thanks to remarkable advances in neuromonitoring data acquisition and analysis. Studies are needed to validate the accuracy and reliability of these new approaches, and their feasibility and implementation within existing intensive care workflows.

Deep learning-enabled detection of hypoxic-ischemic encephalopathy after cardiac arrest in CT scans: a comparative study of 2D and 3D approaches

Objective

To establish a deep learning model for the detection of hypoxic-ischemic encephalopathy (HIE) features on CT scans and to compare various networks to determine the best input data format.

Methods

168 head CT scans of patients after cardiac arrest were retrospectively identified and classified into two categories: 88 (52.4%) with radiological evidence of severe HIE and 80 (47.6%) without signs of HIE. These images were randomly divided into a training and a test set, and five deep learning models based on based on Densely Connected Convolutional Networks (DenseNet121) were trained and validated using different image input formats (2D and 3D images).

Results

All optimized stacked 2D and 3D networks could detect signs of HIE. The networks based on the data as 2D image data stacks provided the best results (S100: AUC: 94%, ACC: 79%, S50: AUC: 93%, ACC: 79%). We provide visual explainability data for the decision making of our AI model using Gradient-weighted Class Activation Mapping.

Conclusion

Our proof-of-concept deep learning model can accurately identify signs of HIE on CT images. Comparing different 2D- and 3D-based approaches, most promising results were achieved by 2D image stack models. After further clinical validation, a deep learning model of HIE detection based on CT images could be implemented in clinical routine and thus aid clinicians in characterizing imaging data and predicting outcome.

Simple approach to quantify hypoxic-ischemic brain injury severity from computed tomography imaging files after cardiac arrest

BACKGROUND

Grey-white ratio (GWR) can estimate severity of cytotoxic cerebral edema secondary to hypoxic-ischemic brain injury after cardiac arrest and predict progression to death by neurologic criteria (DNC). Current approaches to calculating GWR are not standardized and have variable interrater reliability. We tested if measures of variance of brain density on early computed tomographic (CT) imaging after cardiac arrest could predict DNC.

METHODS

We performed a retrospective cohort study, identifying post-arrest patients treated between 2011 and 2020 at our single center. We extracted demographic data from our registry and Digital Imaging and Communication in Medicine (DICOM) files for each patient's first brain CT. We analyzed slices 15-20 of each DICOM, corresponding to the level of the basal ganglia while accommodating differences in patient anatomy. We extracted pixel arrays and converted the radiodensities to Hounsfield units (HU). To focus on brain tissue densities, we excluded HU > 60 and < 10. We calculated the variance of each patient's HU distribution and the difference between the means of a two-group Gaussian finite mixture model. We compared these novel metrics to existing measures of cerebral edema, then randomly divided our data into 80% training and 20% test sets and used logistic regression to predict DNC.

RESULTS

Of 1,133 included subjects, 457 (40%) were female, mean (standard deviation) age was 58 (16) years, and 115 (10%) progressed to DNC. CTs were obtained a median [interquartile range] of 4.2 [2.8-5.7] hours post-arrest. Our novel measures correlated weakly with GWR. HU variance, but not difference between mixture model means, differed significantly between subjects with and without sulcal or cistern effacement. GWR outperformed our novel measures in predicting progression to DNC with an area under the receiver operating characteristic curve (AUC) of 0.82, compared to HU variance (AUC = 0.73) and the difference between mixture model means (AUC = 0.56).

CONCLUSION

There are differences in the distribution of HU on post-arrest CT in patients with qualitative measures of cerebral edema. Current methods to quantify cerebral edema outperform simple measures of attenuation variance on early brain CT. Further analyses could investigate if these measures of variance, or other distributional characteristics of brain density, have improved predictive performance on brain CTs obtained later in the clinical course or derived from discrete regions of anatomical interest.

Measurement of thalamus and cortical damages in hypoxic ischemic encephalopathy

Background

The thalamic gray-white matter ratios (GWRs) on CT and quantitative suppression ratios (SRs) of background activities on EEG may reflect damages in the thalamus and cerebral hemispheres in patients with hypoxic-ischemic encephalopathy (HIE).

Methods

The inclusion criteria were (1) cardiac arrest patients over the age of 20 years from March 2010 to March 2020, and (2) patients who had both EEG and brain CT within 7 days after cardiac arrest. The thalamic GWRs were semi-quantitatively measured by using the region of interest (ROI). SRs of background were analyzed with the installed software (Persyst® v13) in EEG machine.

Results

175 patients were included among 686 patients with HIE and the thalamic GWRs of 168 patients were successfully measured. 155 patients (89 %) showed poor outcomes. The poor outcome group revealed not only higher SRs, but also lower thalamic GWRs. The thalamic GWRs showed a negative correlation to the SRs (ρ (rho) = -0.36, p < 0.0001 for right side, ρ (rho) = -0.31, p < 0.0001 for left side). The good outcome group showed neither beyond the cut-off values of thalamic GWRs nor SRs [40 % (59/148) VS 0 % (0/20) in right side, p = 0.0005 %, and 28 % (42/148) VS 0 % (0/20) in left side, p = 0.0061].

Conclusion

The thalamic GWRs and SRs may reflect the damage in the thalamus and cerebral hemispheres in patients with HIE. Insults in the thalamocortical circuit (TCC) or the thalamus might be responsible for the poor outcome.

Confounders for prognostic accuracy of neuron-specific enolase after cardiac arrest: A retrospective cohort study

AIM

To evaluate neuron-specific enolase (NSE) thresholds for prediction of neurological outcome after cardiac arrest and to analyze the influence of hemolysis and confounders.

METHODS

Retrospective analysis from a cardiac arrest registry. Determination of NSE serum concentration and hemolysis-index (h-index) 48-96 hours after cardiac arrest. Evaluation of neurological outcome using the Cerebral Performance Category score (CPC) at hospital discharge. Separate analyses considering CPC 1-3 and CPC 1-2 as good neurological outcome. Analysis of specificity and sensitivity for poor and good neurological outcome prediction with and without exclusion of hemolytic samples (h-index larger than 50).

RESULTS

Among 356 survivors three days after cardiac arrest, hemolysis was detected in 28 samples (7.9%). At a threshold of 60 µg/L, NSE predicted poor neurological outcome (CPC 4-5) in all samples with a specificity of 92% (86-95%) and sensitivity of 73% (66-79%). In non-hemolytic samples, specificity was 94% (89-97%) and sensitivity 70% (62-76%). At a threshold of 100 µg/L, specificity was 98% (95-100%, all samples) and 99% (95-100%, non-hemolytic samples), and sensitivity 58% (51-65%) and 55% (47-63%), respectively. Possible confounders for elevated NSE in patients with good neurological outcome were ECMO, malignancies, blood transfusions and acute brain diseases. Nine patients with NSE below 17 µg/L had CPC 5, all had plausible death causes other than hypoxic-ischemic encephalopathy.

CONCLUSIONS

NSE concentrations higher than 100 µg/L predicted poor neurological outcome with high specificity. An NSE less than 17 µg/L indicated absence of severe hypoxic-ischemic encephalopathy. Hemolysis and other confounders need to be considered.

INSTITUTIONAL PROTOCOL NUMBER

The local ethics committee (board name: Ethikkommission der Charité) approved this study by the number: EA2/066/23, approval date: 28th June 2023, study title "'ROSC' - Resuscitation Outcome Study."

Quantitative analysis of early apparent diffusion coefficient values from MRIs for predicting neurological prognosis in survivors of out-of-hospital cardiac arrest: an observational study

BACKGROUND

This study aimed to quantitatively analyse ultra-early brain diffusion-weighted magnetic resonance imaging (DW-MRI) findings to determine the apparent diffusion coefficient (ADC) threshold associated with neurological outcomes in comatose survivors of out-of-hospital cardiac arrest (OHCA).

METHODS

This retrospective study included adult survivors of comatose OHCA who underwent DW-MRI imaging scans using a 3-T MRI scanner within 6 h of the return of spontaneous circulation (ROSC). We investigated the association between neurological outcomes and ADC values obtained through voxel-based analysis on DW-MRI. Additionally, we constructed multivariable logistic regression models with pupillary light reflex (PLR), serum neuron-specific enolase (NSE), and ADC values as independent variables to predict poor neurological outcomes. The primary outcome was poor neurological outcome 6 months after ROSC, determined by the Cerebral Performance Category 3-5.

RESULTS

Overall, 131 patients (26% female) were analysed, of whom 74 (57%) showed poor neurological outcomes. The group with a poor neurological outcome had lower mean whole brain ADC values (739.1 vs. 787.1 × 10-6 mm/s) and higher percentages of voxels with ADC below threshold in all ranges (250-1150) (all P < 0.001). The mean whole brain ADC values (area under the receiver operating characteristic curve [AUC] 0.83) and the percentage of voxels with ADC below 600 (AUC 0.81) had the highest sensitivity of 51% (95% confidence interval [CI] 39.4-63.1; cut-off value ≤ 739.2 × 10-6 mm2/s and > 17.2%, respectively) when the false positive rate (FPR) was 0%. In the multivariable model, which also included PLR, NSE, and mean whole brain ADC values, poor neurological outcome was predicted with the highest accuracy (AUC 0.91; 51% sensitivity). This model showed more accurate prediction and sensitivity at an FPR of 0% than did the combination of PLR and NSE (AUC 0.86; 30% sensitivity; P = 0.03).

CONCLUSIONS

In this cohort study, early voxel-based quantitative ADC analysis after ROSC was associated with poor neurological outcomes 6 months after cardiac arrest. The mean whole brain ADC value demonstrated the highest sensitivity when the FPR was 0%, and including it in the multivariable model improved the prediction of poor neurological outcomes.

Inter-rater agreement between humans and computer in quantitative assessment of computed tomography after cardiac arrest

Background

Head computed tomography (CT) is used to predict neurological outcome after cardiac arrest (CA). The current reference standard includes quantitative image analysis by a neuroradiologist to determine the Gray-White-Matter Ratio (GWR) which is calculated via the manual measurement of radiodensity in different brain regions. Recently, automated analysis methods have been introduced. There is limited data on the Inter-rater agreement of both methods.

Methods

Three blinded human raters (neuroradiologist, neurologist, student) with different levels of clinical experience retrospectively assessed the Gray-White-Matter Ratio (GWR) in head CTs of 95 CA patients. GWR was also quantified by a recently published computer algorithm that uses coregistration with standardized brain spaces to identify regions of interest (ROIs). We calculated intraclass correlation (ICC) for inter-rater agreement between human and computer raters as well as area under the curve (AUC) and sensitivity/specificity for poor outcome prognostication.

Results

Inter-rater agreement on GWR was very good (ICC 0.82–0.84) between all three human raters across different levels of expertise and between the computer algorithm and neuroradiologist (ICC 0.83; 95% CI 0.78–0.88). Despite high overall agreement, we observed considerable, clinically relevant deviations of GWR measurements (up to 0.24) in individual patients. In our cohort, at a GWR threshold of 1.10, this did not lead to any false poor neurological outcome prediction.

Conclusion

Human and computer raters demonstrated high overall agreement in GWR determination in head CTs after CA. The clinically relevant deviations of GWR measurement in individual patients underscore the necessity of additional qualitative evaluation and integration of head CT findings into a multimodal approach to prognostication of neurological outcome after CA.

A pilot study of methods for prediction of poor outcome by head computed tomography after cardiac arrest

INTRODUCTION

In Sweden, head computed tomography (CT) is commonly used for prediction of neurological outcome after cardiac arrest, as recommended by guidelines. We compare the prognostic ability and interrater variability of routine and novel CT methods for prediction of poor outcome.

METHODS

Retrospective study including patients from Swedish sites within the Target Temperature Management after out-of-hospital cardiac arrest trial examined with CT. Original images were assessed by two independent radiologists blinded from clinical data with eye-balling without pre-specified criteria, and with a semi-quantitative assessment. Grey-white-matter ratios (GWR) were quantified using models with 4-20 manually placed regions of interest. Prognostic abilities and interrater variability were calculated for prediction of poor outcome (modified Rankin Scale 4-6 at six months) for early (<24h) and late (≥24h) examinations.

RESULTS

68/106 (64%) of included patients were examined <24h post-arrest. Eye-balling predicted poor outcome with 89-100% specificity and 15-78% sensitivity. GWR <24h predicted neurological outcome with unsatisfactory to satisfactory Area Under the Receiver Operating Characteristics Curve (AUROC: 0.54-0.64). GWR ≥24h yielded very good to excellent AUROC (0.80-0.93). Sensitivities increased >2-3 fold in examinations performed after 24h compared to early examinations. Combining eye-balling with GWR<1.15 predicted poor outcome without false positives with sensitivities remaining acceptable.

CONCLUSION

In our cohort, qualitative and quantitative CT methods predicted poor outcome with high specificity and low to moderate sensitivity. Sensitivity increased relevantly after the first 24 hours after CA. Interrater variability poses a problem and indicates the need to standardise brain CT evaluation to increase the methodś safety.