Shanil Dhanji1,2, Mitchell Lawlor1,2

Purpose: Patients with neuro-ophthalmic presenta- tions often have underlying diseases requiring neuro- imaging to make an accurate diagnosis. However, if performed inappropriately, neuroimaging can cause unintended harm. Effective use of neuroimaging in part involves maximizing pre-test probability. This study informs this assessment by identifying which neuro-ophthalmic presentations have relevant radio- logical findings.
Method: We performed a retrospective audit of all neuroimaging ordered for patients attending the Syd- ney Eye Hospital (SEH), Australia, with a neuro- ophthalmic or glaucoma presentation between June 2011 and 2016.
Results: 839 patients underwent neuroimaging for new neuro-ophthalmic or glaucoma presentations. This included 1042 neuroimaging studies: 596 (57%) computed tomography (CT) and 446 (43%) magnetic resonance imaging (MRI). The commonest patient presentations imaged were blurred vision (25%), dip- lopia (18%) and monocular optic disc swelling (11%). Patient presentations with the highest yield of a rele- vant neuroimaging finding included binocular diplo- pia associated with unilateral ptosis (100%) or a combined unilateral cranial nerve palsy (100%). This was followed by Uhthoff’s phenomenon (75%), monocular blurred vision with a visual acuity less than 6/60 (36%) or associated with unilateral pain on eye movements (23%), visual field defects (30%) and transient binocular visual loss (TBVL) (29%). Presentations with a low yield for a relevant finding were central retinal vein occlusion (CRVO), transient monocular visual loss (TMVL), migraine, flashes and/or floaters and isolated ocular pain at rest.
Conclusions: Neuroimaging outcomes may be opti-
mized by reducing requests in conditions where rel- evant findings are uncommon, and using clinical examination to maximize the pre-test probability of the condition being investigated.