Anatomical VOI-based assessment of the three DESH components in probable idiopathic normal pressure hydrocephalus

To evaluate whether anatomical volume-of-interest (VOI) thresholds derived from a SINPHONI-2 reference cohort could provide standardized component-wise quantification of disproportionately enlarged subarachnoid-space hydrocephalus (DESH) in a separate clinical cohort.

Thresholds were derived from 61 SINPHONI-2 cases using a prespecified mildest-case rule and applied without re-optimization to 33 participants with probable idiopathic normal pressure hydrocephalus (iNPH), 26 with Alzheimer disease, 18 with progressive supranuclear palsy, and 20 normal controls. DESH metric-S summed binary positivity for ventriculomegaly, Sylvian fissure enlargement, and tight high-convexity and medial sulci; DESH metric-I was defined as (Sylvian fissure volume + lateral ventricular volume)/high-convexity and medial sulcal volume. Evans index and callosal angle were evaluated as conventional imaging measures.

For iNPH versus pooled non-iNPH, DESH metric-S had an area under the curve (AUC) of 0.958 (95% CI, 0.921-0.990), sensitivity of 97.0% (84.7-99.5%), and specificity of 92.2% (83.0-96.6%). DESH metric-I had an AUC of 0.995 (95% CI, 0.983-1.000), sensitivity of 100.0% (89.6-100.0%), and specificity of 93.8% (85.0-97.5%).

Anatomical VOI-based measures provided standardized component-wise quantification of DESH morphology. DESH metric-S retained the status of each component, whereas DESH metric-I provided a continuous summary. Because DESH contributed to probable-iNPH classification, the observed discrimination reflects concordance with a DESH-supported classification rather than independent diagnostic accuracy for iNPH. External validation, including cohorts spanning incomplete/non-DESH morphology and treatment-response outcomes, is required before clinical or prognostic use.