Segmentation of kidney images is a crucial aspect of the diagnostic and therapeutic application of medical image technologies for kidney disease. Precise kidney segmentation is of the utmost importance in clinical practice, as it enables medical specialists to diagnose diseases and enhance treatment planning. The manual segmentation of the kidneys is a laborious process that is susceptible to variation among specialists as a result of their characteristics. However, precise segmentation of kidney images is difficult to achieve because defining the boundaries of objects in medical images is difficult. An efficient segmentation method for 2D T2-W MRI images of the human abdomen is presented in this work. In this study, kidney segmentation was conducted on patients diagnosed with chronic kidney disease (CKD). The manuscript presents Cascaded-ResUNet++, a segmentation method for kidneys diagnosed with CKD from MRI images, which derives inspiration from ResUNet++. By incorporating distinct residual blocks such as the compression and excitation block, Atrous Spatial Pyramidal Pooling (ASPP), and the attention block, this model aims to outperform ResUNet, AttentionUNet, UNet, and ResUNet++. The Multi-Criteria Decision Analysis (MCDA) method TOPSIS returns a result of 0.704156, indicating that Cascaded-ResUNet++ exhibits superior performance compared to all other methods. In addition, our proposal is highly adaptable for kidney segmentation on account of its cost-effectiveness.