We aimed to assess the interrater variability in optic nerve sheath diameter (ONSD) measurements among neurology residents.

ONSD measurement has shown promise as a non-invasive method for estimating intracranial pressure (ICP), particularly when invasive methods are not available. Training neurology residents in ONSD measurement could expedite detection of patients with intracranial hypertension.

A database of 13 scans from patients admitted to the Neurocritical Care Unit (NCCU) were analyzed. Two neurology residents independently measured ONSD. Interrater reliability was assessed using interclass correlation coefficient. Segmentation for anatomic structures (right dura, right subarachnoid space, optic nerve, left dura, left subarachnoid space) was performed using V7 software and the overlap between both raters was compared.

Mean ONSDint measurements for Rater 1 were 4.90 ± 0.40 mm and ONSDext were 6.033 ± 0.35 mm, whereas mean ONSDint Rater 2 was 4.45 ± 0.72 mm and ONSDext was 6.32 ± 0.80 mm. ICC for the internal borders was 0.86 (95% CI: 0.70-0.94) and 0.094 (95 % CI: 0.04-0.09) for the external borders. Bland-Altman analysis had mean difference of 0.35 mm (95% -0.67 to 1.37 mm) and 0.34 mm (95 % -2.44 to 1.75 mm) for the internal and external borders respectively. IoU for anatomic structures; right dura 38.66, right subarachnoid space 47.52, optic nerve 79.90, left dura 52.07, left subarachnoid space 89.60.

Training residents for ONSD measurement is feasible and yields reliable interrater agreement, particularly for the internal dural borders of ONSD. Standardized training and the use of machine-learning can reduce variability and improve consistency.