Analysis of macular OCT images using deformable registration

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Optical coherence tomography (OCT) is crucial for retinal pathology investigation.
• Deformable image registration is underdeveloped for OCT, limiting population-based analyses.
• Existing generic registration methods are not readily applicable to OCT images.

Purpose of the Study:

• To present a novel deformable image registration approach specifically designed for macular OCT images.
• To validate the algorithm's effectiveness in aligning retinal layers.
• To demonstrate the potential of this method for population-based studies in retinal diseases.

Main Methods:

• Initial fovea alignment via translation.
• Linear rescaling to align retinal boundaries.
• Deformable registration of retinal layers using 1D radial basis functions.

Main Results:

• The algorithm successfully aligned retinal layers in OCT images from healthy controls and MS patients.
• The proposed method overcomes limitations of generic registration techniques for OCT.
• Demonstrated potential for population-based analysis, including a pilot study on MS volumetric changes.

Conclusions:

• The developed deformable image registration algorithm is effective for macular OCT.
• This advancement enables new population-based analytical techniques for retinal imaging.
• Facilitates a deeper understanding of retinal diseases like multiple sclerosis through advanced imaging analysis.