Kidney Computerized Tomography Image Segmentation Using Deep Learning

Abstract

"Kidney CT imaging is important for diagnosing kidney issues and requires accurate segmentation techniques for efficient analysis. Traditional U-Net architecture, which are often used for this task, have limitations with its skip connections, especially during the downsampling process. This process may result in the loss of critical spatial details, which are needed for accurate segmentation. This could happen because skip connections may not completely capture global contextual information and long-range spatial dependencies. In order to tackle these problems, our research suggests the implementation of the Multi-scale Cross Attention Network (MSCA). This network combines attention processes with the U-Net architecture in order to enhance the results of segmentation"