Artificial Intelligence Can Both Measure and Distinguish Retinal Fluid

Optical coherence tomography (OCT) showing signs of macular degeneration. (Photo by: BSIP/Universal Images Group via Getty Images)
Researchers say the retinal fluid volumes can be used as a biomarker to monitor treatment.

Accurate retinal fluid volume data can be efficiently extracted from optical coherence tomography (OCT) scans by artificial intelligence (AI) algorithms and may assist clinicians in managing neovascular age-related macular degeneration (NV-AMD), according to researchers. A study published in the American Journal of Ophthalmology shows that the precise quantitative measures of intraretinal fluid (IRF), subretinal fluid (SRF) and pigment epithelial detachment (PED) can potentially be used as biomarkers by the algorithm to predict changes in NV-AMD and guide treatment.

The research involved scans from 4 retinal image repository datasets — 24,362 scans of 1095 eyes from the HARBOR study ( Identifier: NCT00891735), 511 of 511 from the AREDS2-10 year follow on ( Identifier: NCT03367767), 4673 of 880 from a real-world study in Belfast, and 1470 of 132 from a real-world study in  Tel-Aviv. AI-based automated detection and quantification of IRF, SRF, and PED volumes was used in populations with NV-AMD. 

The investigation shows large ranges that differed by population at the treatment-naïve stage and rapid mean volume decreases at the anti-VEGF therapy maintenance stage. The researchers found that the most appropriate unit of measurement to encompass the entire range of fluid volumes was the nanoliter because it encompassed the full range of fluid volumes.

“In clinical practice, the ability to extract quantitative metrics of exudative activity from OCT scans represents an important advance over the qualitative descriptions that are normally used, where retinal fluid is typically graded as present versus absent, or severe versus mild,” the study says. The quantitative approach the AI offers avoids problems associated with low intra-grader consistency or inter-grader agreement. The technology provides the capability to evaluate exudative activity more meaningfully between sequential and even distant visits of the same patient as well as improve record-keeping and visualization of exudative activity, the researchers report. 

Importantly, the AI software allows assessments and comparisons to be made separately for IRF, SRF, and PED. This is key as prior academic research not all exudative activity is equal. Visual prognoses and the retreatment decisions based on them can be influenced differently depending on the type of exudative activity. 

Future studies are likely to lead to refined retreatment protocols that may well incorporate more nuanced approaches according to fluid compartment or volume.

Limitations of the study included the absence of treatment-naïve retinal volumes in the AREDS2-10Y dataset and diversity in datasets and treatment regimens.

Disclosure: Several study authors declared affiliations with the biotech and pharmaceutical industries. Please see the original reference for a full list of authors’ disclosures.


Keenan TDL, Chakravarthy U, Loewenstein A, Chew EY, Schmidt-Erfurth U. Automated quantitative assessment of retinal fluid volumes as important biomarkers in neovascular age related macular degeneration. Am J Ophthalmol. Published December 21, 2020. doi:10.1016/ j.ajo.2020.12.012