A new study investigated visual system changes in individuals with chronic mild traumatic brain injury. The findings revealed significant deficits across the primary visual pathway and the potential of a wide array of diagnostic tools—including machine learning—for identifying these impairments.
The prospective, observational, case-control study was conducted from May 2018 to November 2021 at a Level 1 trauma research hospital. The study included 28 individuals with a history of mild traumatic brain injury (TBI) and 28 matched controls. Participants were matched by age and sex and had normal best-corrected visual acuity and fundus examinations. The researchers used the Neurobehavioral Symptom Inventory and measurements of oculomotor function, optical coherence tomography, contrast sensitivity, visual evoked potentials, visual field testing, and magnetic resonance imaging (MRI) in a single visit to assess visual dysfunction.
“Vision problems, including light sensitivity, blurred or double vision, and difficulty reading, affect up to 85% of individuals regardless of injury severity,” the investigators described in their recent JAMA Ophthalmology article. While disruptions in ocular motor function, Humphrey visual field testing, and peripapillary retinal nerve fiber layer (RNFL) thickness have been shown in previous studies, “no single metric has consistently captured the full range of visual deficits or been sufficient to diagnose visual dysfunction in individuals with mild TBI” due to the diversity of causes and symptoms of TBI.
Vision problems or light sensitivity was reported by 64% of mild TBI participants, and 78% exhibited measurable visual system deficits. Mild TBI participants exhibited reduced prism convergence test breakpoints and recovery points, which affected binocular vision. Near point of convergence was marginally affected, with a mean difference of 3.26 cm. Monocular and binocular contrast sensitivity were reduced in the mild TBI group, while 31% of mild TBI participants had peripapillary RNFL thinning in the temporal section that was outside the normal range. Given that contrast sensitivity testing has been shown to be a “crucial indicator of visual dysfunction in mild TBI,” the researchers wrote, “the observed deficits in our study further suggest that this test was a sensitive measure for detecting subtle visual impairments associated with this condition.”
MRI revealed changes in optic nerve size and cortical brain volumes in specific regions, including the occipital lobe. The researchers suspected larger optic nerve sizes were a result of glial scarring as a result of axon loss. Participants with both a larger optic nerve and a thinner RNFL self-reported moderate to severe photophobia, “possibly indicating damage to the photosensitive retinal ganglion cells, as these cells contribute to photophobia due to their light sensitive, non–image-forming nature.”
Machine learning models identified abnormalities in 70% of mild TBI participants, particularly in optic radiations and occipital lobe regions. These models achieved a diagnostic accuracy of 72%.
Ultimately, the researchers determined that a comprehensive battery of assessments is more effective at detecting TBI-related vision problems than single metrics in isolation: “The model incorporating all optic radiations and occipital lobe metrics outperformed both the null model and individual models of optic radiations’ microstructure, macrostructure, and occipital lobe segmentation,” they concluded.
A full list of author disclosures can be found in the published research.
