Screening for Intracranial Hypertension Using Lighting-Invariant Computational Pupillometry: A Single-Center Pilot Study

BackgroundInvasive intracranial pressure (ICP) monitoring is the reference standard for detecting intracranial hypertension but requires neurosurgical expertise and carries procedural risks. Non-invasive methods with high negative predictive value (NPV) could serve as screening tools to rule out dangerous ICP elevations and guide decisions about invasive monitoring. We conducted a pilot evaluation to assess whether a smartphone-based, lighting-invariant quantitative pupillometry platform achieves the high NPV needed for clinical screening. MethodsThis prospective single-center pilot study in neuro-ICU enrolled adults with acute neurological emergencies who underwent serial bilateral pupillometry alongside clinically indicated intraparenchymal ICP monitoring. The primary outcome was diagnostic performance of the Pupil Reactivity (PuRe) Score for identifying intracranial hypertension (ICP [&ge;] 20 mmHg), with emphasis on NPV. Secondary analyses examined PuRe Score correlations with ICP, clinical phenotypes across reactivity strata, ambient lighting invariance, and subgroup performance by Glasgow Coma Scale (GCS). ResultsNineteen patients contributed 731 pupillometry recordings, of which 634 (87%) had concurrent invasive ICP measurements; 112 recordings (18%) showed elevated ICP. PuRe Score exhibited a significant inverse correlation with ICP (Spearman{rho} =- 0.17, p<0.001). At the screening threshold (PuRe [&le;]1.3), sensitivity was 85.7% (95% CI: 78.0-91.0%) and specificity 61.3% (95% CI: 57.1-65.4%), yielding a negative predictive value of 95.2% (95% CI: 92.4-97.0%) with AUC of 0.72 (95% CI: 0.68-0.77). Mean ICP differed significantly across PuRe groups: unreactive pupils (PuRe=0) showed 18.3 {+/-}1.3 mmHg versus 8.2 {+/-} 0.5 mmHg in brisk reactivity (PuRe 3; [&ge;] ANOVA F=24.27, p<0.001). The PuRe Score maintained stable discrimination across ambient lighting conditions (ANOVA p=0.91), whereas traditional constriction metrics showed significant lighting dependence (CAMP p=0.04, DELTA p=0.03). ConclusionsIn this pilot cohort, smartphone-based, lighting-invariant pupillometry demonstrated high negative predictive value for ruling out intracranial hypertension. A PuRe Score above the screening threshold may provide bedside reassurance that dangerous ICP elevation is unlikely, suggesting potential for a two-tier neuromonitoring strategy in which high-frequency non-invasive screening identifies patients who require targeted invasive monitoring. Larger validation studies are needed to confirm these findings.