Automated Measurement of Optic Nerve Sheath Diameter

Overview

Optic Nerve Ultrasound (ONUS) is a promising non-invasive tool for the detection of raised Intracranial Pressure (ICP). Variability in the optimal Optic Nerve Sheath Diameter (ONSD) threshold corresponding to elevated ICP in multiple studies limits the value of ONUS in clinical practice. The investigators goal is to develop and validate an automated image analysis algorithm for standardization of ONSD measurement from ultrasound videos. Patients with acute brain injury requiring invasive ICP monitoring will undergo bedside ONUS, with blinded ONSD measurement by an expert investigator. The image analysis algorithm will then be used to measure ONSD and accuracy determined compared to the "reference standard" expert measurement.

BACKGROUND: A promising tool under investigation for the non-invasive estimation of ICP is Optic Nerve Ultrasound (ONUS). Raised ICP results in distension of the optic nerve sheath (ONS), a continuation of the dura mater. Ocular imaging, performed by clinicians using point-of-care ultrasound machines, can detect ONS distension behind the eye. A study conducted at the University of Michigan identified an ONS Diameter (ONSD) cutoff of \>0.51cm as having 98% sensitivity and 91% specificity for the detection of intracranial hypertension, defined as ICP\>25mmHg. The significant variation seen in the optimal ONSD threshold for identification of high ICP across several studies, however, greatly limits the practical application of this technique at the bedside. Much of this variation in the optimal ONSD threshold is likely related to technique, with variation in the margins used to define the ONS on acquired ultrasound images by different operators. The ONS, visible as a linear hypodense structure behind the eye can vary in its visualized dimension based on the angle and plane of insonation. Automated image analysis may permit standardization of ONSD measurement and thereby minimize interobserver variability. SPECIFIC AIM: The investigators goal is to develop a computer image-analysis algorithm to standardize measurement of the ONSD from ultrasound videos, and to validate against the reference standard of expert manual measurement of ONSD. METHODS: Development of image-analysis algorithm: Videos in DICOM format of prior ONUS studies performed for clinical purposes in the neurointensive care unit will be deidentified and used for initial development and modification of the automated image analysis algorithm, prior to prospective enrollment of subjects for determination of accuracy. Under IRB approval, the investigators will perform Optic Nerve Ultrasound (ONUS) on eligible subjects admitted to the ICU following informed consent of the patient or appropriate surrogate. Optic Nerve Ultrasound and ONSD measurement: These patients will undergo ONUS in the ICU. ONUS is performed with the patient's eye closed, and with a linear array transducer placed on the upper margin of the orbit to obtain a sonographic video clip of the eye, followed by bedside measurement of ONSD 3mm behind the level of the posterior scleral border. Imaging will be performed for both eyes for each patient. The expert investigator performing the ONUS study and manual ONSD measurement will be blinded to the patient's ICP by turning the monitor away from the sonographer from the time of entry into the patient's room until exit. The corresponding ICP from the invasive monitor will be separately documented by the bedside nurse. The ONUS video corresponding to the highest measured ONSD will be submitted in DICOM format for automated image analysis. Equipment: Sonosite M-Turbo point-of-care ultrasound machine and an L25 linear array transducer with an ophthalmic preset. Statistical Analysis: * Simple descriptive statistics including means and standard deviations * Simple Pearson correlation to allow for visual inspection across a range of values * Tukey mean-difference plot (Bland - Altman plot) will be used to assess agreement between the two methods, the ultrasound video analysis of optic nerve sheath diameter and ONSD ultrasound interpretation by a clinician. Limits of agreement will be calculated with standard errors and 95% confidence intervals.