Predicting delayed cerebral ischemia after subarachnoid hemorrhage using physiological time series data.

To develop and validate a prediction model for delayed cerebral ischemia (DCI) after subarachnoid hemorrhage (SAH) using a temporal unsupervised feature engineering approach, demonstrating improved precision over standard features. 488 consecutive SAH admissions from 2006 to 2014 to a tertiary care hospital were included. Models were trained on 80%, while 20% were set aside for validation testing. Baseline information and standard grading scales were evaluated: age, sex, Hunt Hess grade, modified Fisher Scale (mFS), and Glasgow Coma Scale (GCS). An unsupervised approach applying random kernels was used to extract features from physiological time series (systolic and diastolic blood pressure, heart rate, respiratory rate, and oxygen saturation). Classifiers (Partial Least Squares, linear and kernel Support Vector Machines) were trained on feature subsets of the derivation dataset. Models were applied to the validation dataset. The performances of the best classifiers on the validation dataset are reported by feature subset. Standard grading scale (mFS): AUC 0.58. Combined demographics and grading scales: AUC 0.60. Random kernel derived physiologic features: AUC 0.74. Combined baseline and physiologic features with redundant feature reduction: AUC 0.77. Current DCI prediction tools rely on admission imaging and are advantageously simple to employ. However, using an agnostic and computationally inexpensive learning approach for high-frequency physiologic time series data, we demonstrated that our models achieve higher classification accuracy.

Duration of Agitation, Fluctuations of Consciousness, and Associations with Outcome in Patients with Subarachnoid Hemorrhage.

BACKGROUND: Agitation is common after subarachnoid hemorrhage (SAH) and may be independently associated with outcomes. We sought to determine whether the duration of agitation and fluctuating consciousness were also associated with outcomes in patients with SAH. METHODS: We identified all patients with positive Richmond Agitation Sedation Scale (RASS) scores from a prospective observational cohort of patients with SAH from 2011 to 2015. Total duration of agitation was extrapolated for each patient using available RASS scores, and 24-h mean and standard deviation (SD) of RASS scores were calculated for each patient. We also calculated each patient's duration of substantial fluctuation of consciousness, defined as the number of days with 24-h RASS SD > 1. Patients were stratified by 3-month outcome using the modified Rankin scale, and associations with outcome were assessed via logistic regression. RESULTS: There were 98 patients with at least one positive RASS score, with median total duration of agitation 8 h (interquartile range [IQR] 4-18), and median duration of substantially fluctuating consciousness 2 days (IQR 1-3). Unfavorable 3-month outcome was significantly associated with a longer duration of fluctuating consciousness (odds ratio [OR] per day, 1.51; 95% confidence interval [CI], 1.04-2.20; p = 0.031), but a briefer duration of agitation (OR per hour, 0.94; 95% CI, 0.89-0.99; p = 0.031). CONCLUSION: Though a longer duration of fluctuating consciousness was associated with worse outcomes in our cohort, total duration of agitation was not, and may have had the opposite effect. Our findings should therefore challenge the intensity with which agitation is often treated in SAH patients.

Prognostication of long-term outcomes after subarachnoid hemorrhage: The FRESH score.

OBJECTIVE: To create a multidimensional tool to prognosticate long-term functional, cognitive, and quality of life outcomes after spontaneous subarachnoid hemorrhage (SAH) using data up to 48 hours after admission. METHODS: Data were prospectively collected for 1,619 consecutive patients enrolled in the SAH outcome project July 1996 to March 2014. Linear models (LMs) were applied to identify factors associated with outcome in 1,526 patients with complete data. Twelve-month functional, cognitive, and quality of life outcomes were measured using the modified Rankin scale (mRS), Telephone Interview for Cognitive Status, and Sickness Impact Profile. Based on the LM residuals, we constructed the FRESH score (Functional Recovery Expected after Subarachnoid Hemorrhage). Score performance, discrimination, and internal validity were tested using the area under the receiver operating characteristic curve (AUC), Nagelkerke and Cox/Snell R(2) , and bootstrapping. For external validation, we used a control population of SAH patients from the CONSCIOUS-1 study (n = 413). RESULTS: The FRESH score was composed of Hunt & Hess and APACHE-II physiologic scores on admission, age, and aneurysmal rebleed within 48 hours. Separate scores to prognosticate 1-year cognition (FRESH-cog) and quality of life (FRESH-quol) were developed controlling for education and premorbid disability. Poor functional outcome (mRS = 4-6) for score levels 1 through 9 respectively was present in 3, 6, 12, 38, 61, 83, 92, 98, and 100% at 1-year follow-up. Performance of FRESH (AUC = 0.90), FRESH-cog (AUC = 0.80), and FRESH-quol (AUC = 0.78) was high. External validation of our cohort using mRS as endpoint showed satisfactory results (AUC = 0.77). To allow for convenient score calculation, we built a smartphone app available for free download. INTERPRETATION: FRESH is the first clinical tool to prognosticate long-term outcome after spontaneous SAH in a multidimensional manner. Ann Neurol 2016;80:46-58.

Bedside quantitative electroencephalography improves assessment of consciousness in comatose subarachnoid hemorrhage patients.

OBJECTIVE: Accurate behavioral assessments of consciousness carry tremendous significance in guiding management, but are extremely challenging in acutely brain-injured patients. We evaluated whether electroencephalography (EEG) and multimodality monitoring parameters may facilitate assessment of consciousness in patients with subarachnoid hemorrhage. METHODS: A retrospective analysis was performed of 83 consecutively treated adults with subarachnoid hemorrhage. All patients were initially comatose and had invasive brain monitoring placed. Behavioral assessments were performed during daily interruption of sedation and categorized into 3 groups based on their best examination as (1) comatose, (2) arousable (eye opening or attending toward a stimulus), and (3) aware (command following). EEG features included spectral power and complexity measures. Comparisons were made using bootstrapping methods and partial least squares regression. RESULTS: We identified 389 artifact-free EEG clips following behavioral assessments. Increasing central gamma, posterior alpha, and diffuse theta-delta oscillations differentiated patients who were arousable from those in coma. Command following was characterized by a further increase in central gamma and posterior alpha, as well as an increase in alpha permutation entropy. These EEG features together with basic neurological examinations (eg, pupillary light reflex) contributed heavily to a linear model predicting behavioral state, whereas brain physiology measures (eg, brain oxygenation), structural injury, and clinical course added less. INTERPRETATION: EEG measures of behavioral states provide distinctive signatures that complement behavioral assessments of patients with subarachnoid hemorrhage shortly after the injury. Our data support the hypothesis that impaired connectivity of cortex with both central thalamus and basal forebrain underlies decreasing levels of consciousness. Ann Neurol 2016;80:541-553.

Agitation After Subarachnoid Hemorrhage: A Frequent Omen of Hospital Complications Associated with Worse Outcomes.

BACKGROUND: Agitated delirium is frequent following acute brain injury, but data are limited in patients with subarachnoid hemorrhage (SAH). We examined incidence, risk factors, and consequences of agitation in these patients in a single-center retrospective study. METHODS: We identified all patients treated with antipsychotics or dexmedetomidine from a prospective observational cohort of patients with spontaneous SAH. Agitation was confirmed by chart review. Outcomes were assessed at 12 months using the modified Rankin Scale (mRS), Telephone Interview for Cognitive Status (TICS), and Lawton IADL (Instrumental Activities of Daily Living) scores. Independent predictors were identified using logistic regression. RESULTS: From 309 SAH patients admitted between January 2011 and December 2015, 52 (17 %) developed agitation, frequently in the first 72 h (50 %) and in patients with Hunt-Hess grades 3-4 (12 % of grades 1-2, 28 % of grades 3-4, 8 % of grade 5). There was also a significant association between agitation and a history of cocaine use or prior psychiatric diagnosis. Agitated patients were more likely to develop multiple hospital complications; and in half of these patients, complications were diagnosed within 24 h of agitation onset. Agitation was associated with IADL impairment at 12 months (Lawton >8; p = 0.03, OR 2.7, 95 % CI, 1.1-6.8) in non-comatose patients (Hunt-Hess 1-4), but not with functional outcome (mRS >3), cognitive impairment (TICS ≤30), or ICU/hospital length of stay after controlling for other predictors. CONCLUSION: Agitation occurs frequently after SAH, especially in non-comatose patients with higher clinical grades. It is associated with the development of multiple hospital complications and may have an independent impact on long-term outcomes.

Oscillatory recruitment of bilateral visual cortex during spatial attention to competing rhythmic inputs.

Selective attention uses temporal regularity of relevant inputs to bias the phase of ongoing population-level neuronal oscillations. This phase entrainment streamlines processing, allowing attended information to arrive at moments of high neural excitability. How entrainment resolves competition between spatially segregated inputs during visuospatial tasks is not yet established. Using high-density electroencephalography in humans, a bilateral entrainment response to the rhythm (1.3 or 1.5 Hz) of an attended stimulation stream was observed, concurrent with a considerably weaker contralateral entrainment to a competing rhythm. That ipsilateral visual areas strongly entrained to the attended stimulus is notable because competitive inputs to these regions were being driven at an entirely different rhythm. Strong modulations of phase locking and weak modulations of single-trial power suggest that entrainment was primarily driven by phase-alignment of ongoing oscillatory activity. In addition, interhemispheric differences in entrained phase were found to be modulated by attended hemifield, implying that the bilateral nature of the response reflected a functional flow of information between hemispheres. This modulation was strongest at the third of at least four harmonics that were strongly entrained. Ipsilateral increases in alpha-band (8-12 Hz) power were also observed during bilateral entrainment, reflecting suppression of the ignored stimulation stream. Furthermore, both entrainment and alpha lateralization significantly affected task performance. We conclude that oscillatory entrainment is a functionally relevant mechanism that synchronizes endogenous activity across the cortical hierarchy to resolve spatial competition. We further speculate that concurrent suppression of ignored input might facilitate the widespread propagation of attended information during spatial attention.

Severe multisensory speech integration deficits in high-functioning school-aged children with Autism Spectrum Disorder (ASD) and their resolution during early adolescence.

Under noisy listening conditions, visualizing a speaker's articulations substantially improves speech intelligibility. This multisensory speech integration ability is crucial to effective communication, and the appropriate development of this capacity greatly impacts a child's ability to successfully navigate educational and social settings. Research shows that multisensory integration abilities continue developing late into childhood. The primary aim here was to track the development of these abilities in children with autism, since multisensory deficits are increasingly recognized as a component of the autism spectrum disorder (ASD) phenotype. The abilities of high-functioning ASD children (n = 84) to integrate seen and heard speech were assessed cross-sectionally, while environmental noise levels were systematically manipulated, comparing them with age-matched neurotypical children (n = 142). Severe integration deficits were uncovered in ASD, which were increasingly pronounced as background noise increased. These deficits were evident in school-aged ASD children (5-12 year olds), but were fully ameliorated in ASD children entering adolescence (13-15 year olds). The severity of multisensory deficits uncovered has important implications for educators and clinicians working in ASD. We consider the observation that the multisensory speech system recovers substantially in adolescence as an indication that it is likely amenable to intervention during earlier childhood, with potentially profound implications for the development of social communication abilities in ASD children.

Frontal networks associated with command following after hemorrhagic stroke.

BACKGROUND AND PURPOSE: Level of consciousness is frequently assessed by command-following ability in the clinical setting. However, it is unclear what brain circuits are needed to follow commands. We sought to determine what networks differentiate command following from noncommand following patients after hemorrhagic stroke. METHODS: Structural MRI, resting-state functional MRI, and electroencephalography were performed on 25 awake and unresponsive patients with acute intracerebral and subarachnoid hemorrhage. Structural injury was assessed via volumetric T1-weighted MRI analysis. Functional connectivity differences were analyzed against a template of standard resting-state networks. The default mode network (DMN) and the task-positive network were investigated using seed-based functional connectivity. Networks were interrogated by pairwise coherence of electroencephalograph leads in regions of interest defined by functional MRI. RESULTS: Functional imaging of unresponsive patients identified significant differences in 6 of 16 standard resting-state networks. Significant voxels were found in premotor cortex, dorsal anterior cingulate gyrus, and supplementary motor area. Direct interrogation of the DMN and task-positive network revealed loss of connectivity between the DMN and the orbitofrontal cortex and new connections between the task-positive network and DMN. Coherence between electrodes corresponding to right executive network and visual networks was also decreased in unresponsive patients. CONCLUSIONS: Resting-state functional MRI and electroencephalography coherence data support a model in which multiple, chiefly frontal networks are required for command following. Loss of DMN anticorrelation with task-positive network may reflect a loss of inhibitory control of the DMN by motor-executive regions. Frontal networks should thus be a target for future investigations into the mechanism of responsiveness in the intensive care unit environment.

Interests shape how adolescents pay attention: the interaction of motivation and top-down attentional processes in biasing sensory activations to anticipated events.

The voluntary allocation of attention to environmental inputs is a crucial mechanism of healthy cognitive functioning, and is probably influenced by an observer's level of interest in a stimulus. For example, an individual who is passionate about soccer but bored by botany will obviously be more attentive at a soccer match than an orchid show. The influence of monetary rewards on attention has been examined, but the impact of more common motivating factors (i.e. the level of interest in the materials under observation) remains unclear, especially during development. Here, stimulus sets were designed based on survey measures of the level of interest of adolescent participants in several item classes. High-density electroencephalography was recorded during a cued spatial attention task in which stimuli of high or low interest were presented in separate blocks. The motivational impact on performance of a spatial attention task was assessed, along with event-related potential measures of anticipatory top-down attention. As predicted, performance was improved for the spatial target detection of high interest items. Further, the impact of motivation was observed in parieto-occipital processes associated with anticipatory top-down spatial attention. The anticipatory activity over these regions was also increased for high vs. low interest stimuli, irrespective of the direction of spatial attention. The results also showed stronger anticipatory attentional and motivational modulations over the right vs. left parieto-occipital cortex. These data suggest that motivation enhances top-down attentional processes, and can independently shape activations in sensory regions in anticipation of events. They also suggest that attentional functions across hemispheres may not fully mature until late adolescence.

Modulation of early cortical processing during divided attention to non-contiguous locations.

We often face the challenge of simultaneously attending to multiple non-contiguous regions of space. There is ongoing debate as to how spatial attention is divided under these situations. Whereas, for several years, the predominant view was that humans could divide the attentional spotlight, several recent studies argue in favor of a unitary spotlight that rhythmically samples relevant locations. Here, this issue was addressed by the use of high-density electrophysiology in concert with the multifocal m-sequence technique to examine visual evoked responses to multiple simultaneous streams of stimulation. Concurrently, we assayed the topographic distribution of alpha-band oscillatory mechanisms, a measure of attentional suppression. Participants performed a difficult detection task that required simultaneous attention to two stimuli in contiguous (undivided) or non-contiguous parts of space. In the undivided condition, the classic pattern of attentional modulation was observed, with increased amplitude of the early visual evoked response and increased alpha amplitude ipsilateral to the attended hemifield. For the divided condition, early visual responses to attended stimuli were also enhanced, and the observed multifocal topographic distribution of alpha suppression was in line with the divided attention hypothesis. These results support the existence of divided attentional spotlights, providing evidence that the corresponding modulation occurs during initial sensory processing time-frames in hierarchically early visual regions, and that suppressive mechanisms of visual attention selectively target distracter locations during divided spatial attention.

Oscillatory alpha-band suppression mechanisms during the rapid attentional shifts required to perform an anti-saccade task.

Neuroimaging has demonstrated anatomical overlap between covert and overt attention systems, although behavioral and electrophysiological studies have suggested that the two systems do not rely on entirely identical circuits or mechanisms. In a parallel line of research, topographically-specific modulations of alpha-band power (~8-14 Hz) have been consistently correlated with anticipatory states during tasks requiring covert attention shifts. These tasks, however, typically employ cue-target-interval paradigms where attentional processes are examined across relatively protracted periods of time and not at the rapid timescales implicated during overt attention tasks. The anti-saccade task, where one must first covertly attend for a peripheral target, before executing a rapid overt attention shift (i.e. a saccade) to the opposite side of space, is particularly well-suited for examining the rapid dynamics of overt attentional deployments. Here, we asked whether alpha-band oscillatory mechanisms would also be associated with these very rapid overt shifts, potentially representing a common neural mechanism across overt and covert attention systems. High-density electroencephalography in conjunction with infra-red eye-tracking was recorded while participants engaged in both pro- and anti-saccade task blocks. Alpha power, time-locked to saccade onset, showed three distinct phases of significantly lateralized topographic shifts, all occurring within a period of less than 1s, closely reflecting the temporal dynamics of anti-saccade performance. Only two such phases were observed during the pro-saccade task. These data point to substantially more rapid temporal dynamics of alpha-band suppressive mechanisms than previously established, and implicate oscillatory alpha-band activity as a common mechanism across both overt and covert attentional deployments.

Atypical cortical representation of peripheral visual space in children with an autism spectrum disorder.

A key feature of early visual cortical regions is that they contain discretely organized retinotopic maps. Titration of these maps must occur through experience, and the fidelity of their spatial tuning will depend on the consistency and accuracy of the eye movement system. Anomalies in fixation patterns and the ballistics of eye movements are well documented in autism spectrum disorder (ASD), with off-center fixations a hallmark of the phenotype. We hypothesized that these atypicalities might affect the development of visuo-spatial maps and specifically that peripheral inputs might receive altered processing in ASD. Using high-density recordings of visual evoked potentials (VEPs) and a novel system-identification approach known as VESPA (visual evoked spread spectrum analysis), we assessed sensory responses to centrally and peripherally presented stimuli. Additionally, input luminance was varied to bias responsiveness to the magnocellular system, given previous suggestions of magnocellular-specific deficits in ASD. Participants were 22 ASD children (7-17 years of age) and 31 age- and performance-IQ-matched neurotypical controls. Both VEP and VESPA responses to central presentations were indistinguishable between groups. In contrast, peripheral presentations resulted in significantly greater early VEP and VESPA amplitudes in the ASD cohort. We found no evidence that anomalous enhancement was restricted to magnocellular-biased responses. The extent of peripheral response enhancement was related to the severity of stereotyped behaviors and restricted interests, cardinal symptoms of ASD. The current results point to differential visuo-spatial cortical mapping in ASD, shedding light on the consequences of peculiarities in gaze and stereotyped visual behaviors often reported by clinicians working with this population.

Early spatial attentional modulation of inputs to the fovea.

Attending to a specific spatial location modulates responsivity of neurons with receptive fields processing that part of the environment. A major outstanding question is whether attentional modulation operates differently for the foveal (central) representation of the visual field than it does for the periphery. Indeed, recent animal electrophysiological recordings suggest that attention differentially affects spatial integration for central and peripheral receptive fields in primary visual cortex. In human electroencephalographic recordings, spatial attention to peripheral locations robustly modulates activity in early visual regions, but it has been claimed that this mechanism does not operate in foveal vision. Here, however, we show clear early attentional modulation of foveal stimulation with the same timing and cortical sources as seen for peripheral stimuli, demonstrating that attentional gain control operates similarly across the entire field of view. These results imply that covertly attending away from the center of gaze, which is a common paradigm in behavioral and electrophysiological studies of attention, results in a precisely timed push-pull mechanism. While the amplitude of the initial response to stimulation at attended peripheral locations is significantly increased beginning at 80 ms, the amplitude of the response to foveal stimulation begins to be attenuated.

Inferior pedicle autoaugmentation mastopexy after breast implant removal.

BACKGROUND: A new method of autoaugmentation mammaplasty is presented to correct ptosis and to increase the projection and volume of the breast in patients who would like a reposition augmentation mammaplasty after breast implant removal but do not want a new implant. METHODS: Between 1999 and 2007, a total of 27 patients (age = 54 +/- 7.3 years) underwent mammaplasty using an inferior-based flap of deepithelialized subcutaneous and breast tissue modularized to its pedicle which was inserted beneath a superior pedicle used for correction of ptosis and to increase the projection and apparent volume of the breast. RESULTS: The results confirmed that autoaugmentation mammaplasty of the breast following removal of the implant yields longstanding results. It corrects ptosis and increases the projection and apparent volume of the breast when mastopexy is planned without use of a new implant. Twelve months after surgery the degree of descent of the inframammary fold generally parallels that of the nipple. The mean level of the inframammary fold was below the mean level of the nipple. Postoperatively, the optimum distance had been largely achieved. CONCLUSION: The advantages of the technique presented here are that it minimizes the skin scar in cases using vertical mammaplasty techniques and optimizes the breast shape after breast implant removal in patients who do not want a new implant.

Atypical processing of tones and phonemes in Rett Syndrome as biomarkers of disease progression.

Due to severe motor impairments and the lack of expressive language abilities seen in most patients with Rett Syndrome (RTT), it has proven extremely difficult to obtain accurate measures of auditory processing capabilities in this population. Here, we examined early auditory cortical processing of pure tones and more complex phonemes in females with Rett Syndrome (RTT), by recording high-density auditory evoked potentials (AEP), which allow for objective evaluation of the timing and severity of processing deficits along the auditory processing hierarchy. We compared AEPs of 12 females with RTT to those of 21 typically developing (TD) peers aged 4-21 years, interrogating the first four major components of the AEP (P1: 60-90 ms; N1: 100-130 ms; P2: 135-165 ms; and N2: 245-275 ms). Atypicalities were evident in RTT at the initial stage of processing. Whereas the P1 showed increased amplitude to phonemic inputs relative to tones in TD participants, this modulation by stimulus complexity was absent in RTT. Interestingly, the subsequent N1 did not differ between groups, whereas the following P2 was hugely diminished in RTT, regardless of stimulus complexity. The N2 was similarly smaller in RTT and did not differ as a function of stimulus type. The P2 effect was remarkably robust in differentiating between groups with near perfect separation between the two groups despite the wide age range of our samples. Given this robustness, along with the observation that P2 amplitude was significantly associated with RTT symptom severity, the P2 has the potential to serve as a monitoring, treatment response, or even surrogate endpoint biomarker. Compellingly, the reduction of P2 in patients with RTT mimics findings in animal models of RTT, providing a translational bridge between pre-clinical and human research.

Autoaugmentation mastopexy with an inferior-based pedicle.

Mammaplasty for breast enhancement and correction of ptosis augmentation is described. Between 2002 and 2007, autoaugmentation mammaplasty was performed for 27 patients (age, 48 +/- 7.3 years) using an inferior-based flap of deepithelialized dermoglandular tissue inserted beneath the breast parenchyma of a superior-based nipple-areolar complex pedicle. The results confirmed that autoaugmentation mammaplasty corrects ptosis while increasing the projection and apparent volume of the breast. The degree of inframammary fold (IMF) descent 6 months after surgery generally paralleled that of the nipple. The mean level of the IMF was below the mean level of the nipple. Postoperatively, the optimum distance had been largely achieved. The advantage of the technique is that it optimizes the shape and volume of the breast without the use of an implant.

A phase II randomized controlled trial of tiopronin for aneurysmal subarachnoid hemorrhage.

OBJECTIVE: Delayed cerebral ischemia (DCI) is a significant contributor to poor outcomes after aneurysmal subarachnoid hemorrhage (aSAH). The neurotoxin 3-aminopropanal (3-AP) is upregulated in cerebral ischemia. This phase II clinical trial evaluated the efficacy of tiopronin in reducing CSF 3-AP levels in patients with aSAH. METHODS: In this prospective, randomized, double-blind, placebo-controlled, multicenter clinical trial, 60 patients were assigned to receive tiopronin or placebo in a 1:1 ratio. Treatment was commenced within 96 hours after aSAH onset, administered at a dose of 3 g daily, and continued until 14 days after aSAH or hospital discharge, whichever occurred earlier. The primary efficacy outcome was the CSF 3-AP level at 7 ± 1 days after aSAH. RESULTS: Of the 60 enrolled patients, 29 (97%) and 27 (93%) in the tiopronin and placebo arms, respectively, received more than one dose of the study drug or placebo. At post-aSAH day 7 ± 1, CSF samples were available in 41% (n = 12/29) and 48% (n = 13/27) of patients in the tiopronin and placebo arms, respectively. No difference in CSF 3-AP levels at post-aSAH day 7 ± 1 was observed between the study arms (11 ± 12 nmol/mL vs 13 ± 18 nmol/mL; p = 0.766). Prespecified adverse events led to early treatment cessation for 4 patients in the tiopronin arm and 2 in the placebo arm. CONCLUSIONS: The power of this study was affected by missing data. Therefore, the authors could not establish or refute an effect of tiopronin on CSF 3-AP levels. Additional observational studies investigating the role of 3-AP as a biomarker for DCI may be warranted prior to its use as a molecular target in future clinical trials.Clinical trial registration no.: NCT01095731 (ClinicalTrials.gov).

What's color got to do with it? The influence of color on visual attention in different categories.

Certain locations attract human gaze in natural visual scenes. Are there measurable features, which distinguish these locations from others? While there has been extensive research on luminance-defined features, only few studies have examined the influence of color on overt attention. In this study, we addressed this question by presenting color-calibrated stimuli and analyzing color features that are known to be relevant for the responses of LGN neurons. We recorded eye movements of 15 human subjects freely viewing colored and grayscale images of seven different categories. All images were also analyzed by the saliency map model (L. Itti, C. Koch, & E. Niebur, 1998). We find that human fixation locations differ between colored and grayscale versions of the same image much more than predicted by the saliency map. Examining the influence of various color features on overt attention, we find two extreme categories: while in rainforest images all color features are salient, none is salient in fractals. In all other categories, color features are selectively salient. This shows that the influence of color on overt attention depends on the type of image. Also, it is crucial to analyze neurophysiologically relevant color features for quantifying the influence of color on attention.

Incorporating High-Frequency Physiologic Data Using Computational Dictionary Learning Improves Prediction of Delayed Cerebral Ischemia Compared to Existing Methods.

PURPOSE: Accurate prediction of delayed cerebral ischemia (DCI) after subarachnoid hemorrhage (SAH) can be critical for planning interventions to prevent poor neurological outcome. This paper presents a model using convolution dictionary learning to extract features from physiological data available from bedside monitors. We develop and validate a prediction model for DCI after SAH, demonstrating improved precision over standard methods alone. METHODS: 488 consecutive SAH admissions from 2006 to 2014 to a tertiary care hospital were included. Models were trained on 80%, while 20% were set aside for validation testing. Modified Fisher Scale was considered the standard grading scale in clinical use; baseline features also analyzed included age, sex, Hunt-Hess, and Glasgow Coma Scales. An unsupervised approach using convolution dictionary learning was used to extract features from physiological time series (systolic blood pressure and diastolic blood pressure, heart rate, respiratory rate, and oxygen saturation). Classifiers (partial least squares and linear and kernel support vector machines) were trained on feature subsets of the derivation dataset. Models were applied to the validation dataset. RESULTS: The performances of the best classifiers on the validation dataset are reported by feature subset. Standard grading scale (mFS): AUC 0.54. Combined demographics and grading scales (baseline features): AUC 0.63. Kernel derived physiologic features: AUC 0.66. Combined baseline and physiologic features with redundant feature reduction: AUC 0.71 on derivation dataset and 0.78 on validation dataset. CONCLUSION: Current DCI prediction tools rely on admission imaging and are advantageously simple to employ. However, using an agnostic and computationally inexpensive learning approach for high-frequency physiologic time series data, we demonstrated that we could incorporate individual physiologic data to achieve higher classification accuracy.