# Comparative CT Ventricular Morphometrics in Hydrocephalus, Stroke, and Traumatic Brain Injury: A Distortion-Controlled Analysis

## TL;DR

This study shows that CT scans can help distinguish hydrocephalus from stroke and brain injury by measuring ventricular size and shape.

## Contribution

The study introduces a distortion-controlled CT morphometric model that improves diagnostic accuracy for hydrocephalus.

## Key findings

- Hydrocephalus patients had significantly larger ventricles compared to stroke and TBI patients.
- Adjusting for intracranial distortion increased model accuracy to an AUC of 0.91.
- Evans index and third ventricle width were the strongest predictors of hydrocephalus.

## Abstract

Background/Objectives: Ventricular enlargement is a common finding on non-contrast computed tomography (CT) in acute neurological presentations, occurring in hydrocephalus, stroke, and traumatic brain injury. This study evaluated whether routinely available CT-based ventricular morphometric parameters can distinguish hydrocephalus from stroke and traumatic brain injury using initial imaging examinations. Methods: This retrospective observational study included 186 adults (68 with hydrocephalus, 64 with stroke, and 54 with TBI) who underwent index non-contrast cranial CT. Quantitative ventricular parameters included Evans index and third ventricle width, alongside temporal horn dilation and periventricular edema. Multivariable logistic regression models were developed to assess diagnostic performance. A primary morphometric model and a distortion-controlled model incorporating midline shift, mass lesions, and hemorrhage burden were analyzed. Discrimination was evaluated using receiver operating characteristic (ROC) curves. Results: Patients with hydrocephalus showed significantly greater ventricular enlargement, with higher Evans index and third ventricle width compared with stroke and traumatic brain injury groups. The primary model demonstrated moderate discrimination (AUC 0.71). After adjustment for intracranial distortion variables, model performance improved substantially (AUC 0.91), with balanced sensitivity and specificity at optimized thresholds. Evans index and third ventricle width were the strongest independent predictors. Conclusions: CT-derived ventricular morphometrics provide a practical and reproducible approach for differentiating hydrocephalus from stroke and traumatic brain injury on first-presentation CT, supporting objective interpretation in routine neuroimaging practice.