IterMiUnet: A lightweight architecture for automatic blood vessel segmentation

The automatic segmentation of blood vessels in fundus images can help analyze the condition retinal vasculature, which is crucial for identifying various systemic diseases like hypertension, diabetes, etc. Despite success Deep Learning-based models this task, most them are heavily parametrized and thus have limited use practical applications. This paper proposes IterMiUnet, a new lightweight convolution-based model that requires significantly fewer parameters yet delivers performance similar to existing models. makes excellent capabilities Iternet architecture but overcomes its nature by incorporating encoder-decoder structure MiUnet within it. Thus, reduces without any compromise with network's depth, necessary learn abstract hierarchical concepts deep speeds up training inference time potentially helpful medical domain where data scarce and, therefore, tend overfit. proposed was evaluated on three publicly available datasets: DRIVE, STARE, CHASE-DB1. Further cross-training inter-rater variability evaluations also been performed. has lot potential be utilized as tool early diagnosis many diseases.