Enhanced Image Segmentation in Medical Imaging Using Mini-Net

Computer-aided diagnosis and analysis of medical imaging depend on the precise segmentation of anatomical structures and anomalies. High segmentation accuracy is attained by Deep Learning (DL) algorithms; nevertheless, real-time applications are hindered by their computing needs. Furthermore, a lot of advanced segmentation techniques might not be the best for medical imaging even while they work well for segmenting objects in general. In order to improve the segmentation of medical images, this study presents Mini net, a lightweight segmentation network. Mini-Net can analyse data in real time because it efficiently records both high- and low-frequency information with less than 38,000 parameters. MoNuSeg, ISIC-2016, ISIC-2018, and DRIVE are among the benchmark datasets used to assess Mini-Net. The results show its durability and strength. Mini-Net achieves competitive accuracy compared to existing state-of-the-art methods while maintaining efficiency.