For cardiac arrest (CA) survivors who are initially comatose, once circulation has been reestablished, the extent of brain injury and expected neurologic outcome is a decisive factor for decisions regarding long-term management. Poor neurologic prognosis commonly leads to withdrawal of life-sustaining therapy (WLST) and subsequent death. Prognostic techniques have traditionally relied on the clinical examination,¹ electrophysiological measurements^{2, 3} or biochemical changes.⁴ Unfortunately, the most rigorous studies of the prognostic value of the clinical examination were performed decades ago, 1 and since that time there have been significant advances in treatment options, such as targeted temperature management (TTM).⁵ Traditional recommendations for neurologic prognostication have proven unreliable in modern studies of CA patients,^{6, 7} motivating the search for advanced techniques that can better interrogate the degree of cerebral injury and likelihood of recovery. Several studies have demonstrated that default mode network (DMN) connectivity is altered in patients recovering from coma.⁸ The thalamus has long been considered a key player in consciousness as a relay center and modulator of peripheral sensory information to the cortex,⁹ and is central to attention, sleep-wake state and arousal.¹⁰ We hypothesize that patients with poor outcomes will exhibit thalamocortical functional network (TCN) and DMN disruptions compared to those who wake-up.CA patients who were comatose for at least 24hours while not being cooled were prospectively enrolled. Coma was defined as Glasgow Coma Scale (GCS) <=8. All subjects underwent 3T MRI. Resting state functional MRI (rs-fMRI) was acquired using 150 gradient echo echo-planar imaging measurements, field-of-view (FOV) of 220x220 mm², 64x64 acquisition matrix, 3 mm thick skip of 0.5 mm and temporal resolution of 2400 msec. High-spatial resolution 3D T1-weighted multi-echo magnetization prepared gradient echo (MEMPRAGE) anatomical images were acquired for registration purposes with FOV=256x256 mm², acquisition matrix=256x256, 176 sagittal slices (thickness 1 mm), 3xGRAPPA acceleration. Functional connectivity analyses will be performed using a modified version of previously published pipelines (NITRC fcon 1000 script).¹¹ Images were slice time corrected (FSL), motion-corrected, spatial filtered, and co-registered to the MNI152 T1 2mm brain and nuisance signals (e.g., global signal, white matter, motion parameters) regressed out. Seeds were either based on the posterior cingulate cortex (PCC) seed distributed as part of fcon¹¹ and resampled to the MNI152 T1 2mm brain atlas or bilateral thalamic regions calculated from the Harvard-Oxford Sub-cortical Structural Atlas¹² using a 50% probability threshold. Correlation coefficients were calculated on a voxel-wise basis with respect to either the PCC or thalamic seeds and transformed to Z-scores. DMN and TCN maps from all subjects were compared to those from 4 healthy controls using nonparametric permutation testing (N=5000).¹³ DMN and TCN maps from patients with good outcome (eye opening spontaneously or to stimuli) were compared to those with poor outcomes (failure to recover arousal before discharge).¹³ CA patients’ mean age (±SD) was 46.4±26.4; 70% were female. Duration of arrest was known in 6 patients (33.8±19.7 min). Seven patients exhibited eye opening. Seven patients had died as a result of WLST. Admission GCS was 3. At the time of MRI, median [IQR] GCS was 6.5 [3-7.25]. Median [IQR] time of the research MRI was 6 [3-8.75] days. Figure 1 shows the DMN for healthy controls, patients with recovery of arousal and patients without recovery. Shown are 1-P-values. Also shown are comparisons between healthy controls vs all patients, as well as between awake and non-awake patients. Significant differences are observed between healthy controls and patients especially with respect to the medial prefrontal cortex. Differences between awake and non-awake patients DMN appear to be primarily cerebellar. Figure 2 shows the results of similar analysis for the TCN, for which disruptions in the TCN are readily apparent. Differences between awake and non-awake patients were found primarily within the pons and regions of the cerebellum.Patients who failed to regain consciousness demonstrated greater disturbances in both DMN and TCN resting-state network. These findings suggest that rs-fMRI may have utility in identifying patients who may have good outcome despite presenting with poor GCS scores. Differences in timing of MRI acquisition and potential bias from self-fulfilling prophecy are limitations of our findings. Although the research rs-fMRI results were not made available to the clinical team, the other clinical MRI data that were shared may have influenced treatment decisions. Future prospective studies are needed for which decisions regarding withdrawal of care are deferred for at least two weeks post-arrest in order to accurately characterize patient’s likelihood for recovery.