Related Experiment Video
Updated: Jan 4, 2026
10:10
Ultrahigh Resolution Mouse Optical Coherence Tomography to Aid Intraocular Injection in Retinal Gene Therapy Research
Published on: November 2, 2018
9.7K
Inter-scanner Variation Independent Descriptors for Constrained Diffeomorphic Demons Registration of Retina OCT
S Reaungamornrat1, A Carass1, Y He1
1Department of Neurology, Johns Hopkins Hospital, Baltimore, MD.
Summary
We developed D-MIND Demons, a novel deformable image registration method for Optical Coherence Tomography (OCT) scans. This robust technique improves accuracy in retinal imaging studies, overcoming image quality limitations and inter-scanner variability.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Ophthalmology
Background:
- Optical Coherence Tomography (OCT) provides high-resolution retinal imaging for studying neurodegenerative and ocular diseases.
- Accurate inter-scanner deformable image registration is crucial for longitudinal OCT studies.
- OCT image quality is often suboptimal due to noise and poor through-plane resolution, complicating registration.
Purpose of the Study:
- To develop a robust deformable image registration method for OCT scans that addresses image quality issues and inter-scanner variability.
- To incorporate constraints for improperly defined OCT geometry.
- To create a method insensitive to noise and other image degradations.
Main Methods:
- A diffeomorphic deformable registration method, D-MIND Demons, was developed using decentralized-modality-insensitive-neighborhood-descriptors (D-MINDs).
- The method incorporates constraints on velocity fields within a MIND-Demons framework.
- D-MINDs were evaluated for descriptiveness against other descriptors, and D-MIND Demons performance was compared for cross-scanner, intra-, and inter-subject registration using clinical OCT data.
Main Results:
- D-MINDs demonstrated superior descriptiveness compared to other descriptors.
- D-MIND Demons proved computationally efficient and robust against OCT image degradation (noise, speckle, intensity non-uniformity, poor through-plane resolution).
- Consistent registration accuracy was achieved (4±4 μm cross-scanner intra-subject, 4±6 μm cross-scanner inter-subject) irrespective of denoising.
Conclusions:
- The developed D-MIND Demons method offers a promising solution for OCT image registration in ophthalmological and neurological research.
- This approach can enhance image segmentation, disease progression evaluation, and patient population analysis.
- Facilitates improved diagnosis and personalized treatment strategies through accurate retinal structure analysis.
