Related Experiment Video
Updated: Dec 14, 2025
08:32
Optimized Protocol for Retinal Wholemount Preparation for Imaging and Immunohistochemistry
Published on: December 13, 2013
23.9K
Forming Optimal Projection Images from Intra-Retinal Layers Using Curvelet-Based Image Fusion Method
Jalil Jalili1,2, Hossein Rabbani2, Alireza Mehri Dehnavi2
1Medical Physics and Biomedical Engineering Unit, Ophthalmic Research Center, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Journal of Medical Signals and Sensors
|July 18, 2020
Summary
Curvelet-based image fusion enhances projection images from optical coherence tomography (OCT), revealing detailed retinal structures. This method surpasses average and wavelet techniques for superior image fusion and feature extraction.
Area of Science:
- Ophthalmology
- Medical Imaging
- Biomedical Engineering
Background:
- Image fusion combines information from multiple images.
- Three-dimensional (3D) optical coherence tomography (OCT) reveals retinal pathology not seen in fundus images.
- Current average-based projection methods lose intraretinal details.
Purpose of the Study:
- To develop an optimal projection image formation method using Curvelet-based image fusion.
- To enhance the visualization of intraretinal details and features from 3D OCT data.
Main Methods:
- Segmentation of 3D OCT macular spectral data into 11 retinal layers.
- Generation of projection images using statistical methods between retinal layer boundaries.
- Application of Curvelet transform for merging retinal layers and creating final projection images.
Main Results:
- Curvelet-based fused images integrate retinal depth information.
- Enhanced extraction of retinal features like vessels and the macula region.
- Superior qualitative and quantitative performance compared to average-based and wavelet-based fusion methods, achieving high entropy (6.7744) and AG (9.5491).
Conclusions:
- Curvelet-based image fusion produces images with significantly higher contrast and more detailed information.
- Thin retinal veins, often absent in other methods, are clearly visible in Curvelet-based fused images.
