Computer Assisted Analysis of Ocular Imaging

Enhancement, Segmentation, Registration, Fusion and Classification

Adaptive-Weighted Bilateral Filtering and Other Pre-processing Techniques for Optical Coherence Tomography (CMIG, 2014)

Aim

This project aims to develop advanced image analysis tools to maximise the information inherent in retinal images. This comprehensive approach encompasses several key methodologies: enhancement to improve image quality, segmentation to isolate critical structures, registration to align images from different modalities, fusion to integrate complementary data, and classification to diagnose and monitor ocular conditions. The focus is on leveraging these techniques across various imaging modalities, including color fundus photography, optical coherence tomography (OCT), and confocal microscopy. By integrating these tools, the project seeks to provide a robust framework for the detailed analysis and interpretation of retinal images, facilitating early detection, diagnosis, and treatment of ocular diseases. This multidisciplinary effort aims to contribute significantly to the field of ophthalmology, offering enhanced diagnostic capabilities and ultimately improving patient outcomes.

Collaborators
School of Optometry and Vision Sciences (Cardiff University)
Funder
China Scholarship Council (CSC)

Research team

Methods

  • Texture-based glaucoma classification (ISIB2013)

    An automated texture classification method for glaucoma was developed, utilising robust principal component analysis of texture descriptors. A multi-modal information fusion technique included visual field measurements, OCT, and retinal fundus photography.

  • Adaptive-Weighted Bilateral Filtering (ICIP2013)

    The method removes speckle while preserving useful retinal layer information using multi-scale despeckling based on a dual-tree complex wavelet transform (DT-CWT). Further enhancement is achieved through a novel adaptive-weighted bilateral filter (AWBF) that preserves texture.

Downloads

Publications
Datasets
Related publications from VI-Lab
Detection in different modalities