Loss of sight is prevalent among the elderly subjects. Various eye diseases, such as age-related macular degeneration, diabetic retinopathy and glaucoma are the major causes of loss of vision. Early detection of such diseases can impede the progression of blindness. However, it is time consuming and laborious to conduct mass screening. Hence, computer aided diagnosis can help to overcome these drawbacks.
The technology includes a decision support system that can discriminate between the four classes automatically using Pyramid Histogram of Words (PHOW). The extracted features from PHOW are used to build the vocabulary and encoded using Fisher Vector to train the classification model. This system can differentiate the four classes with an accuracy of 96.79%, sensitivity of 96.73% and specificity of 96.96%.
This technology can be applied to the healthcare industry for eye screening, to support ophthalmologists in diagnosing abnormal fundus images. It can reduce the workload of the ophthalmologists during eye screening hence, increasing productivity and reducing screening time for patients. The technology can be extended to detect other eye diseases like diabetes maculopathy, floaters, retinal detachment and macular hole.
Faster screening time and less workload for ophthalmologists.