"Imaging the Image" - Enhancing Image Characteristics Of Retinal Images Of Retinopathy Of Prematurity Using An Indigenous Patent Pending Software ("RetiView")

Abstract

 

Purpose:

 

To report the utility of an indigenously developed signal processing based visualization software ("RetiView") in detecting poorly visible and undetected features in retinal images of infants with Retinopathy of Prematurity (ROP).

 

Methods:

 

"RetiView" uses Anisotropic Diffusion Filtering (ADF) and Contrast Limited Histogram Equalization (CLAHE) techniques for retinal image enhancement of noisy, low contrast retinal images. ‘Vesselness measure’ is calculated and a non-linear pseudo coloring algorithm is used for visualization and analysis of vascular structures. Three protocols were developed: ‘Grey Enhancement (GE)’, ‘Color Enhancement (CE)’ and ‘Vesselness Measure (VNM)’. Vascular analysis of wide-field digital images of infant retinas with ROP obtained on the Retcam Shuttle was performed using ‘multiscale vessel enhancement filtering’ techniques. Image enhancement of poorly discernible, clinically relevant features were analyzed in images of aggressive posterior ROP (APROP) and atypical ROP.

 

Results:

 

CE protocol allowed better visualization of capillary non-perfusion areas and enhanced the edge of vascular loops in APROP. GE protocols were used to study neovascularization. The VNM protocol was useful in determining the extent and distribution of the vessels including enhancing traceability upto the ora serrata. Both protocols helped in delineating zones of avascularity, edge of zone 1 vessels and differentiating between vascular and fibrotic components within the extra-retinal fibrovascular proliferation.

 

Conclusions:

 

This pilot study demonstrates the successful feasibility of this indigenous software which enhances the visualization of clinically undetectable changes in atypical images of ROP non-invasively.  

 

Keywords: image processing • imaging/image analysis: clinical • retina