Automatic detection of true retinal area and diagnosing retinal disease using SLO images

Abstract: Artificial Neural Network (ANN) classifier can be wont to detection of retinal diseases. Scanning laser ophthalmoscopes (SLOs) will be used for early detection of retinal diseases. It is a technique of examination of the attention. The advantage of using SLO is its wide field of read, which will image an oversized a part of the tissue layer for higher identification of the retinal diseases. On the other facet, during the imaging method, artefacts such as eyelashes and eyelids also are imaged together with the retinal area. This brings a big challenge on the way to exclude these artefacts. In proposed novel approach to mechanically extract out true retinal space from An SLO image based mostly on image process and machine learning approaches. The Simple Linear Iterative Clustering (SLIC) is that the rule utilized in super-pixel calculation. To decrease the unpredictability of image getting ready errands and provide an advantageous primitive picture style. To reduce the complexness of image process tasks and supply a convenient primitive image pattern, also to classified pixels into completely different regions based mostly on the regional size and compactness, known as super-pixels. The framework then calculates image based options reflective textural data and classifies between retinal space and artefacts. The experimental evaluation results have shown sensible performance with a high accuracy.

Keywords: Feature selection, retinal artefacts extraction, retinal image analysis, scanning laser ophthalmoscope (SLO) Superpixel Classification.