Conflicting Central Subfield Thickness in a Cataractous, Highly Myopic Eye
A 68-year-old man is referred for baseline OCT before cataract surgery. He has a dense nuclear sclerotic cataract (NO4), axial length 29.6 mm, type 2 diabetes (last HbA1c 7.8%), and best-corrected acuity of 20/80. The first macular cube (512×128) reports a central subfield thickness of 344 µm and flags 'intraretinal fluid'; signal strength is 4/10. A repeat cube ten minutes later reports 296 µm at signal strength 5/10. The ordering physician asks whether the patient has diabetic macular edema that should be treated before surgery. Reviewing the B-scans, the technician sees the inner boundary drawn along a hyperreflective line several pixels above the true internal limiting membrane temporally, and the RPE line breaks into two segments beneath the fovea. No cystoid spaces are visible on the raw images.
1. How do you decide whether the 344 µm versus 296 µm difference reflects true pathology or measurement artifact?
Read the worked answer
Start from the principle that automated thickness is a derivative product: the number is only as good as the boundaries. A 48 µm difference between two scans exceeds typical test–retest variability for a well-segmented macula, so at least one (possibly both) values are unreliable.
Work through the B-scans, not the report:
- Check each boundary layer by layer. The ILM line sits above the true surface temporally and the RPE line is discontinuous under the fovea. Both errors corrupt the inner and outer limits of the measurement.
- Ask whether the flagged 'fluid' is real. True intraretinal fluid appears as discrete hyporeflective spaces with defined walls; shadowing from a NO4 cataract and low signal produce broad pseudohyporeflective bands that the algorithm can mislabel.
- Consider signal strength. At 4–5/10, automated segmentation is below the threshold generally accepted for reliable output (approximately 6/10 or better).
- Exclude other artifacts. High myopia produces a curved retinal contour that can fall outside the scan window and generate mirror or truncation artifact; check the en face image for centration and the scan window for cut-off edges.
- Look for corroboration. Is there a prior OCT? Does acuity, fundoscopy, or the fellow eye fit edema?
Conclusion: two discordant values with visible segmentation failure and no cysts on the raw images do not establish edema; the measurement is inconclusive.
2. What acquisition and workflow changes would you make to obtain a reliable measurement, and what would you communicate to the physician?
Read the worked answer
Acquisition changes
- Dilate if the pupil is small, and ensure the tear film is intact (have the patient blink immediately before capture; consider lubricant).
- Optimize focus and clean the objective; confirm the patient is positioned with the chin on the rest and forehead against the strap.
- Switch to a denser/faster protocol or an averaged scan to raise signal, and repeat the cube; use the follow-up/rescan mode to compare the same location.
- If signal remains poor because of the cataract, consider a device with a longer acquisition wavelength or enhanced-depth imaging, or defer until after cataract surgery if clinically acceptable.
- Manually inspect and, if the software permits, correct the segmentation before the report is finalized.
Communication
State plainly: 'Signal strength 4–5/10 with visible ILM and RPE segmentation errors; central thickness values of 344 and 296 µm are not reliable, and no cystoid spaces are seen on the raw B-scans. Recommend repeat OCT after cataract surgery or with a higher-quality acquisition before initiating anti-VEGF therapy.' Flag the scan as quality-limited in the report rather than leaving an unflagged abnormal thickness value.