Two-Stage Segmentation and Ensemble Modeling: Kidney Tumor Analysis in CT Images

In the realm of kidney cancer, accurate segmentation is pivotal for effective diagnosis and treatment. Participating in the 2023 KiTS Challenge as a platform, our research introduces a two-stage strategy combining the strengths of nnU-Net and nnFormer for enhanced tumor segmentation. Our approach focuses on the kidney region, facilitating the learning of tumor-influenced areas, and employs an ensemble of two nnU-Net models for precise segmentation. Evaluated on the KiTS23 dataset, which emphasizes the segmentation of the kidney, tumor, and cyst, our method demonstrated its potential in addressing complex medical image segmentation challenges.