This is the work I did in my third year summer internship at the University of Warwick under the guidance of Under Dr. Abhir Bhalerao
, Dr S. S. Anand
. The work got published in a IEEE conference and also got the best paper special mention award 🙂 The work was completed in 10 weeks time. Warwick was a nice serene place to work in and the research visit was amazing.
Abstract from the paper -
The feature space
Diabetic retinopathy is one of the major causes of blindness. However, diabetic retinopathy does not usually cause a loss of sight until it has reached an advanced stage. The earliest sign of the disease are microaneurysms (MA) which appear as small red dots on retinal fundus images. Various screening programmes have been established in the UK and other countries to collect and assess images on a regular basis, especially in the diabetic population. A considerable amount of time and money is spent in manually grading these images, a large percentage of which are normal. By automatically identifying the normal images, the manual workload and costs could be reduced greatly while increasing the effectiveness of the screening programmes. A novel method of microaneurysm detection from digital retinal screening images is proposed. It is based on filtering using a complex-valued circular-symmetric filters, and an eigen-image, morphological analysis of the candidate regions to reduce the false-positve rate. We detail the image processing algorithms and present results on a typical set of 89 image from a published database. Our method is shown to have a best operating sensitivity of 82.6% at a specificity of 80.2% which makes it viable for screening. We discuss the results in the context of a model of visual search and the ROC curves that it can predict.
The paper can be downloaded from IEEE explore
if you have a subscription.
Print This Post