Machine-learning based segmentation of the optic nerve head using multi-contrast Jones matrix optical coherence...

Area of Science:

• Biomedical Optics
• Medical Imaging
• Ophthalmology

Background:

• Optical Coherence Tomography (OCT) is a key imaging modality in ophthalmology.
• Limitations exist in standard OCT for detailed tissue characterization.
• Jones matrix OCT (JM-OCT) offers multi-contrast imaging capabilities.

Purpose of the Study:

• To develop and demonstrate a pixel-by-pixel tissue classification framework using JM-OCT.
• To enable detailed analysis of optic nerve head substructures.
• To facilitate advanced optical property analysis of specific tissues.

Main Methods:

• Utilized Jones matrix optical coherence tomography (JM-OCT) for multi-contrast data acquisition.
• Developed a classification framework involving feature engineering and k-means clustering.
• Trained a tissue classifier on synthetic features derived from JM-OCT contrasts.
• Applied the framework for in vivo analysis of posterior eye optic nerve heads.

Main Results:

• Successfully classified JM-OCT pixels into prelamina, lamina cribrosa (lamina beam), and retrolamina tissues.
• Demonstrated pixel-level tissue differentiation in optic nerve heads.
• Enabled subsequent birefringence and attenuation coefficient analysis of the lamina beam.

Conclusions:

• The JM-OCT based tissue classification framework is effective for optic nerve head analysis.
• This approach enhances the diagnostic capabilities of OCT in ophthalmology.
• The methodology allows for detailed investigation of ocular tissue optical properties.