Assessing quality of OCT scans impacts clinical-decision process

July 7, 2019

With imaging artifacts common, clinicians must be focused on key points

Many ocular factors contribute to the incidence of artifacts. Myopic patients are not represented in the normative database even though it is an important cause of errors or misinterpretation.

Reviewed by David S. Greenfield, MD

An assessment of the quality of optical coherence tomography (OCT) scans can impact clinical decisions related to the diagnosis of glaucoma and its progression, according to David S. Greenfield, MD. Because imaging artifacts are common, found in as many as 20% of nerve fiber layer scans and almost 30% of scans involving the macular region, particularly in those patients that have coexisting macular pucker, clinicians should pay careful attention to all aspects of their scans, said Dr. Greenfield, Douglas R. Anderson distinguished professor of Ophthalmology and vice chair for academic affairs at the Bascom Palmer Eye Institute, University of Miami Miller School of Medicine.

A number of ocular factors contribute to the incidence of artifacts, a frequent one being that highly myopic patients are not represented in the normative database despite the fact that high myopia is known to be an important cause of errors or misinterpretation, Dr. Greenfield said. Other factors to watch for include significant ocular surface disease, opacification in the lens or vitreous, and peripapillary atrophy.

Four parts of the scan to study closely are signal strength (refer to the manufacturer’s specifications); the thickness map (look for the presence or absence of shadowing artifacts); the deviation map (has the optic disk been automatically and accurately delineated, and can any subtle features related to motion artifacts be detected); and individual tomograms (search for signs of segmentation failure).

Monitoring the recommended individual signal-to-noise ratios is clinically relevant because a reduction in the signal strength will not only affect the estimation of retinal fiber layer (RNFL) thickness, but is also highly associated with an increase in the measurement’s variability and failure of the measurement algorithm (as demonstrated in the figure).

Details in the imaging

RNFL imaging artifacts can be very subtle, arising from causes such as a horizontal translation of the blood vessels due to eye movement during imaging acquisition. Other common problems are blink artifacts and shadowing artifacts from vitreous opacities.

Eyes with severe pathologic myopia often have what Dr. Greenfield refers to as the “trifecta,” an incorrect axial alignment with shadowing artifacts, poor delineation of the optic disc border, and clear evidence of segmentation failure, all contributing to RNFL measurements that are “off the chart.”

Macular artifacts are even more common than RNFL artifacts and can be found in eyes with pathology on the surface of the macula, intraretinal edema, and even associated with subretinal pathology such as choroidal neovascularization, confounding the GCIPL measurement algorithm. Extremely large regions of macular pathology can also adversely affect the RNFL thickness algorithm and resemble glaucomatous RNFL atrophy.

RELATED: Understanding advantages, limitations of SS-OCT“But that does not mean that eyes with comorbid conditions should not also undergo testing in other anatomic regions,” Dr. Greenfield added, referring to the case of a patient with cystoid macular edema due to a macular pucker.

The GCIPL thickness had been increasing over time due to worsening CME, masking the ability to detect progressive glaucoma. Serial examination of the RNFL images showed progressive atrophy in the average, superior, and inferior RNFL thickness that was not detectable in the macular region because of the presence of CME.

“Poor quality scans with some measures may not preclude high quality scans with other measures in different anatomic regions,” Dr. Greenfield said.

In terms of disease detection, many normal eyes with high axial myopia have anomalous tilted discs making it difficult to clinically establish the presence of glaucomatous damage, he noted. Close examination may show that the neural rim of the optic disc is intact, and the visual field may show a characteristic enlargement of the physiologic blind spot, he added.

The false identification of glaucoma using OCT in such eyes has been referred to as “red disease” (RNFL thickness values outside of the normal range that are displayed in red lettering or boxes). The thickness values in such eyes are flagged as statistically abnormal simply because they are poorly represented in the normative database. Some eyes may have early glaucomatous damage despite normal RUNFL imaging, often referred to as “green disease.”

Dr. Greenfield managed the case of a glaucoma patient with a reproducible paracentral visual field scotoma, normal RNFL thickness, and localized atrophy in GCIPL thickness due to centralized glaucoma damage in the macular region. Such cases remind clinicians of the importance of examining the central 10-2 visual field and GCIPL thickness in addition to performing RNFL imaging in eyes with glaucoma, he said.

A final factor that must be considered when analyzing OCT scans over time is the impact of the aging process, Dr. Greenfield said, adding that serial imaging over an extended period of time may demonstrate slow changes in the RNFL thickness due to normal aging. He urged caution when interpreting eyes with very slowly progressive, isolated changes in RNFL thickness with stable IOP and stable visual function.

RELATED: Mini Quiz: Educating patients about DME

Disclosures:

David S. Greenfield, MD
P: 516/515-1500 E: d.greenfield@med.miami.edu
This article was adapted from Dr. Greenfield’s presentation at the 2019 meeting of the American Glaucoma Society. He did not report any relevant financial disclosures.