Extending the classical approach of Bags of Visual Words, the latest RECOD research on Diabetic Retinopathy – Beyond Lesion-based Diabetic Retinopathy: a Direct Approach for Referral – has just been published on the Journal of Biomedical and Health Informatics (J-BHI), which is among the top 3 (out of 69) Health Information Management Journals according to H-index. The paper results from an international cooperation between the researchers of RECOD Lab and Prof. Herbert Jelinek (Charles Sturt University, Australia).
Here’s the abstract:
Diabetic retinopathy (DR) is the leading cause of blindness in adults, but can be managed if detected early. Automated DR screening helps by indicating which patients should be referred to the doctor. However, current techniques of automated screening still depend too much on the detection of individual lesions. In this work we bypass lesion detection, and directly train a classifier for DR referral. Additional novelties are the use of state-of-the-art mid-level features for the retinal images: BossaNova and Fisher Vector. Those features extend the classical Bags of Visual Words and greatly improve the accuracy of complex classification tasks. The proposed technique for direct referral is promising, achieving an area under the curve (AUC) of 96.4%, thus reducing the classification error by almost 40% over the current state of the art, held by lesion-based techniques.
Ramon Pires, Sandra Avila, Herbert F. Jelinek, Jacques Wainer, Eduardo Valle and Anderson Rocha. Beyond Lesion-based Diabetic Retinopathy: a Direct Approach for Referral. Journal of Biomedical and Health Informatics (J-BHI), 2015. 10.1109/JBHI.2015.2498104