• COVID-19
  • Biosimilars
  • Cataract Therapeutics
  • DME
  • Gene Therapy
  • Workplace
  • Ptosis
  • Optic Relief
  • Imaging
  • Geographic Atrophy
  • AMD
  • Presbyopia
  • Ocular Surface Disease
  • Practice Management
  • Pediatrics
  • Surgery
  • Therapeutics
  • Optometry
  • Retina
  • Cataract
  • Pharmacy
  • IOL
  • Dry Eye
  • Understanding Antibiotic Resistance
  • Refractive
  • Cornea
  • Glaucoma
  • OCT
  • Ocular Allergy
  • Clinical Diagnosis
  • Technology

How to detect artifactual clues in OCT images


Achieve more accurate findings by eliminating the effects of artifacts

Take-Home Message

Optical coherence tomography is indispensable for diagnosing glaucoma. However, clinicians should be alert to the effects of various artifacts.


By Lynda Charters; Reviewed by Sanjay G. Asrani, MD

Durham, NC-The ability to identify various types of artifacts is crucial when evaluating optical coherence tomography (OCT) images.

OCT has become indispensable for diagnosing glaucoma. However, as with all technologies, there are glitches that can be overcome with proper recognition, explained Sanjay G. Asrani, MD.

“It is critical to recognize artifacts in images, because many ophthalmologists rely on imaging devices to identify glaucoma and detect disease progression,” said Dr. Asrani, professor of ophthalmology, and director, Glaucoma OCT Reading Center, Department of Ophthalmology, Duke Eye Center, Duke University, Durham, NC. Among the types of artifacts:

Operator-dependent artifacts

Operator-dependent artifacts include acceptance of images with poor signal strength, improper retinal nerve fiber layer (RNFL) circle placement, and truncation of acquired OCT images. Dr. Asrani provided an example of a scan that showed complete absence of the RNFL in one sector, which is highly unusual because of the floor effect of the glial cells.

“This finding indicated the presence of an artifact,” he said. “When the scan was repeated, the RNFL appeared to be nearly normal. This happened as the result of truncation because the technician failed to bring the image into the center of the window.”

Misinterpretation artifacts

Misinterpretation artifacts include localized losses of RNFL or macular thickness classified as normal due to averaging of the thickness values by quadrant or hemisphere and misinterpretation of shadow artifacts.

He provided an example that he referred to as “green disease,” in which the green color of the data results provides false assurance that the findings are normal. In this case, the patient had a focal defect that was easily missed by averaging the quadrant data. A second example was that of a shadow artifact caused by a floater that resulted in borderline measurements.

Ocular pathology-related artifacts

Ocular pathology-related artifacts include the presence of prominent posterior hyaloid and epiretinal membrane (ERM) that create abnormal hyperreflective bands inside the normal retinal boundary. In this situation, the algorithm may identify these pathologies as the retinal boundary, which can result in overestimation of the retinal thickness.

Dr. Asrani showed images depicting a falsely thick RNFL and another of a falsely thick macula, both due to the ERM. In the first example, the clue to the artifact was that the axons appeared to be hanging from the ERM in an area that appeared thickened but was not.

In the second example, scalloped edges around an artificially thick macula were clues to the artifactual effect of the ERM.

In patients with posterior vitreous detachment (PVD), an artifact can result from operator failure to recognize the effect of a partial PVD that causes traction-related thickening of the peripapillary RNFL.

“This is one of the most common artifacts seen,” Dr. Asrani emphasized.

A related artifact that is very difficult to detect occurs in cases that seem to have PVD progression and subsequent thinning of the RNFL. However, unless the vitreous interface and the RNFL edges are evaluated, the physician cannot detect whether the vitreous has pulled away from the RNFL, which then relaxes and appears as false progression, he noted.

Myopia, which he referred to as “red disease,” presents difficulties. There is a higher percentage of abnormal diagnostic classifications, because the RNFL normative databases usually do not include patients with moderate and high myopia. Myopia is also associated with other artifacts (i.e., difficulty acquiring a good image due to the long axial length or myopic retinal schisis that affects the peripapillary RNFL thickness).

Dr. Asrani described a case in which peripapillary atrophy caused substantial measurement artifacts and another case in which measurements were not obtained because of axial length, posterior staphyloma, and retinal schisis.

Software-relate artifacts

Software-related artifacts include absence of eye tracking because of microsaccades, not controlling for head tilt, and the absence of robust segmentation software.

Failure to recognize non-glaucomatous diseases-such as anterior ischemic optic neuropathy, retinal dystrophies, and hemiretinal vein occlusion, among others-is a source of another set of artifacts.

Ocular pathologies, specifically ERM and vitreous traction, caused the most artifacts in macular and RNFL scans, he noted.

“OCT has tremendous potential for glaucoma management, but we need to be aware of artifacts,” Dr. Asrani concluded. “Re-acquiring scans or manually correcting segmentation errors, while time-consuming, may be needed for better clinical care. Continued improvements in software and segmentation algorithms may provide increasingly reliable quantitative thickness data.”

Sanjay G. Asrani, MD

P: 919/684-8656

E: sanjay.asrani.duke.edu

Dr. Asrani receives lecture fees from Heidelberg Engineering. This article is adapted from Dr. Asrani’s presentation during Glaucoma 2012 at the 2012 annual meeting of the American Academy of Ophthalmology.

Related Videos
© 2024 MJH Life Sciences

All rights reserved.