Fast segmentation of anterior segment optical coherence tomography images using graph cut

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Optical coherence tomography (OCT) provides non-invasive anterior segment imaging.
• Accurate segmentation of OCT images is crucial for patient-specific biomechanical eye models.
• These models can enhance ophthalmic diagnosis and treatment planning.

Purpose of the Study:

• To develop a novel, automated segmentation technique for OCT images of the anterior segment.
• To evaluate the accuracy and efficiency of the new segmentation method.
• To compare the technique against manual segmentation and existing methods.

Main Methods:

• A graph cut algorithm incorporating regional and shape terms was developed.
• The technique was applied to segment 39 anterior segment OCT images.
• Performance was assessed using Dice's similarity coefficient (DSC), MSPE, and Hausdorff distance, compared to manual and level set methods.

Main Results:

• The novel technique achieved a mean DSC of 0.943 ± 0.020, outperforming previous methods.
• Significant reductions in processing time were observed.
• The method demonstrated high accuracy and efficiency in segmenting OCT images.

Conclusions:

• A fast and accurate graph cut-based segmentation technique for OCT images was successfully developed.
• This method holds potential for improving diagnostic accuracy and treatment planning in ophthalmology.
• Automated segmentation facilitates the creation of patient-specific biomechanical eye models.