Automatic microaneurysm detection of diabetic retinopathy in fundus images

Diabetic retinopathy (DR) is a serious diabetic complication, and Microaneurysm (MA) is the earliest lesion in diabetic retinopathy, so early MA detection plays a critical role in diabetic retinopathy diagnosis. In this paper, we propose the Joint Dynamic Sparse Representation (JDSR) algorithm with multiple-channel multiple-feature dictionaries. Candidates for MA are first extracted as small image blocks; then we develop the multiple-channel multiple-feature dictionaries for candidate representation. Next, sparse coefficient can be obtained by the proposed JDSR algorithm which can be used for classification. Additionally, in order to form an optimal dictionary, the group sparsity dictionary selection method is also introduced. We evaluate our algorithm by comparing it with other state-of-the-art algorithms. Extensive experiment results on ROC database demonstrate the effectiveness of the proposed algorithm.