Contrast sensitivity is a key metric of visual function that strongly correlates with visual impairment and daily functional vision. Patients with age-related macular degeneration (AMD) often complain of visual dysfunction in the early stages of their disease, but their visual acuity is typically spared. Current assessments of contrast sensitivity better correlate with functional vision during normal daily activities than with visual acuity.

In clinical research, contrast sensitivity function (CSF) has demonstrated the potential to help monitor the progression of eye diseases like AMD. However, the absence of a reliable and clinically applicable test has hindered adoption of CSF testing as a routine practice, says Neal S. Patel, MD, from Massachusetts Eye and Ear of Harvard University.

In recent years, different types of artificial intelligence and deep learning tools that use ocular imaging data have been developed to enhance ophthalmologic care. Applying these tools to vision testing may help clinicians get a clearer picture of contrast sensitivity and acuity, which in turn could improve clinical decision making and patient outcomes.

New Research

Dr. Patel and a team of investigators presented data at the 2020 American Academy of Ophthalmology Annual Meeting that characterized CSF in patients with non-neovascular AMD (nnAMD) using a novel computerized active learning device. The study group developed the Quick Contrast Sensitivity Function (qCSF) method, which combines artificial intelligence with active learning to make the measurement process quicker and more accurate. The model draws from 128 possible contrasts and 19 spatial frequencies, amounting to more than 2,400 candidate test items.

 In the study, researchers measured CSF using the qCSF with the Manifold Contrast Vision Meter (Adaptive Sensory Technology; San Diego, CA) in 129 eyes of patients with nnAMD and compared them with a healthy control group consisting of 133 eyes. Study outcomes included area under the log CSF (AULCSF), contrast sensitivity thresholds at spatial frequencies ranging from 1 to 18 cycles per degree, and best corrected visual acuity (BCVA).

Key Findings

According to the results, CSF measured with the qCSF test was significantly decreased in early nnAMD. A multivariate regression analysis showed that CSF thresholds at low spatial frequencies were significantly decreased in early nnAMD despite no observed differences in BCVA. For intermediate and advanced nnAMD, CSF thresholds at low spatial frequencies and AULCSF were lower than what was seen in the control group. No significant differences were observed at higher spatial frequencies. Notably, AULCSF was able to differentiate between nnAMD stages.

Dr. Patel notes that the qCSF test had high sensitivity and good test-test-reliability. He says it also took only about 5 minutes per eye to complete and maximized information gained. “CSF may emerge as a promising visual functional endpoint in clinical practice and future nnAMD clinical trials based on the study results,” he says. “Future investigations may include longitudinal studies with qCSF tested over the progression of macular disease.”