Reliability of a computer-aided manual procedure for segmenting optical coherence tomography scans

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Accurate measurement of retinal layer thickness is crucial for diagnosing and monitoring glaucoma.
• Frequency-domain optical coherence tomography (FD-OCT) is a key imaging modality for glaucoma assessment.
• Manual segmentation of FD-OCT scans requires consistent and reliable performance from operators.

Purpose of the Study:

• To evaluate the consistency of manual segmentation of retinal layers in FD-OCT scans.
• To assess both within-operator (test-retest) and between-operator agreement.
• To determine the reliability of computer-aided manual segmentation for glaucoma research.

Main Methods:

• Four trained individuals manually segmented FD-OCT scans from glaucoma patients and healthy controls using a computer-aided tool.
• Scans were segmented twice by each operator with at least a one-month interval to assess within-operator reliability.
• Concordance correlation coefficients and mean absolute differences in border locations were used to quantify agreement.

Main Results:

• Good within-operator reliability was observed, with mean concordance correlations ranging from 0.978 to 0.999.
• Mean differences in border locations (MEAN{mean( ΔLBL )}) ranged from 1.6 to 4.7 micrometers.
• Good between-operator agreement was also found, with concordance correlations between 0.992 and 0.999, and MEAN{mean( ΔLBL )} values from 1.9 to 4.0 micrometers.

Conclusions:

• Trained human segmenters demonstrate high reliability in manual border segmentation of FD-OCT scans.
• Manual segmentation provides a dependable method for measuring retinal layer thickness in glaucoma studies.
• The consistency of this method supports its use in clinical research for glaucoma assessment.