FE-UNet: A Feature-Enhanced Deep Learning Framework for Kidney MRI Segmentation

Accurate kidney segmentation from abdominal magnetic resonance imaging (MRI) is critical for identifying and treating chronic kidney disease (CKD). Nevertheless, manual segmentation is time-consuming and is subject to inter-observer variability. This study introduces FE-UNet, a feature enhanced UNet framework for automatic and reliable kidney segmentation from 2D T2-weighted abdominal MRIs, designed to enhance feature sharing and stability. Here, FE-UNet achieves a Dice Coefficient of 0.8185 outperforming 15 leading variations of UNet. A multi-criteria evaluation strategy TOPSIS together with statistical significance testing and radiological assessment, is employed to comprehensively compare performance across multiple segmentation metrics. It improves kidney MRI segmentation by providing better performance in biomedical imaging.