Automated circumpapillary retinal nerve fiber layer segmentation in high-resolution swept-source OCT

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Glaucoma is a progressive optic neuropathy causing retinal ganglion cell loss and thinning of the circumpapillary retinal nerve fiber layer (RNFL).
• RNFL thickness measurement is crucial for glaucoma diagnosis and monitoring.
• Manual RNFL assessment is time-consuming and prone to subjectivity.

Purpose of the Study:

• To develop and evaluate a deep learning model for automated RNFL segmentation from OCT images.
• To compare the performance of the automated method with manual assessment.

Main Methods:

• Extracted circumpapillary OCT scans from volumetric scans using a high-resolution swept-source OCT device.
• Utilized manual annotations for training and evaluating a deep learning-based segmentation model.
• Assessed the accuracy and diagnostic performance of the automated segmentation.

Main Results:

• The deep learning model achieved automated RNFL segmentation with accuracy comparable to manual assessment.
• The diagnostic performance of the automated method was found to be on par with manual evaluation.
• Demonstrated the potential of deep learning for objective and efficient RNFL analysis.

Conclusions:

• Automated RNFL segmentation using deep learning is a viable alternative to manual measurement.
• This approach offers improved objectivity and efficiency in glaucoma diagnosis and monitoring.
• Deep learning holds significant promise for advancing quantitative imaging analysis in ophthalmology.