Comparison of the Iowa Reference Algorithm to the Heidelberg Spectralis optical coherence tomography segmentation...

Area of Science:

• Ophthalmology
• Neuroscience
• Medical Imaging

Background:

• Accurate retinal layer thickness measurement is crucial for studying neurodegenerative diseases using spectral-domain optical coherence tomography (SD-OCT).
• Segmentation algorithms can introduce variability in measurements and errors, necessitating an understanding of their performance and the impact of manual corrections.

Purpose of the Study:

• To compare retinal layer thickness measurements between two common SD-OCT segmentation algorithms: the Iowa Reference Algorithm and Heidelberg Spectralis.
• To identify locations of segmentation errors and assess the impact of manual correction on thickness measurements in patients with frontotemporal degeneration.

Main Methods:

• Macular SD-OCT images from frontotemporal degeneration patients and controls were analyzed.
• Retinal layer thickness measurements from the Iowa Reference Algorithm and Heidelberg Spectralis were compared.
• Manual correction of significant segmentation errors was performed and its impact evaluated.

Main Results:

• Small differences were observed in most retinal layer thickness measurements between the two algorithms.
• Outer sectors of the Early Treatment Diabetic Retinopathy Study grid showed a higher percentage of eyes requiring correction compared to inner sectors for the retinal nerve fiber layer (RNFL).
• Manual corrections had a mild effect, causing at most a 5% change in RNFL thickness.

Conclusions:

• Researchers can be informed on the optimal use of different segmentation algorithms for comparing retinal layer thicknesses in neurodegeneration studies.
• Understanding algorithm-specific errors and the effect of manual corrections enhances the reliability of SD-OCT in clinical research.