User-guided segmentation for volumetric retinal optical coherence tomography images

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Manual segmentation of retinal layers in optical coherence tomography (OCT) images is time-consuming but essential for accurate measurements.
• Existing automatic segmentation techniques often fail in regions with irregular retinal layers.

Purpose of the Study:

• To develop a user-guided segmentation method for retinal layers and features in OCT images.
• To bridge the gap between manual and automatic segmentation, improving efficiency and accuracy.

Main Methods:

• A user-guided approach where users sketch irregular retinal regions in 3D OCT images.
• Utilizing novel layer and edge detectors based on robust likelihood estimation, guided by user input.
• Tracing entire 3D retinal layers and anatomical features for segmentation.

Main Results:

• The proposed method successfully segments retinal layers and features in both mouse and human OCT images.
• Demonstrated reliability and efficiency compared to purely automatic methods.
• User guidance effectively handles regions where automatic methods fail.

Conclusions:

• The user-guided segmentation method offers a reliable and efficient solution for OCT image analysis.
• This approach enhances the accuracy of retinal layer segmentation, particularly in complex cases.
• It provides a valuable tool for clinical research and diagnostics involving OCT imaging.