Answers are grounded in this paper and cite their sources.
SGA-U-Net shows improved segmentation accuracy for histopathological classification of Wilms tumor, suggesting enhanced clinical application.
Wilms tumor (WT) is the most prevalent renal malignancy in children. Determining its histopathological classification is critical for prognosis and postoperative treatment options. The histopathological classification of WT is based on the area percentages of its primary components, making accurate segmentation of these components essential for classification outcomes. However, due to the complexity of WT components and the high resolution of whole slide images (WSIs), achieving precise pathological diagnosis presents quiet challenges. Hence, we propose a new SGA-U-Net for the segmentation of WT components. To improve the model's focus on fine-grained features within the WT components, a hybrid attention module is designed for the up-sampling layer of the traditional U-Net. We also applied the model to assess the histopathological classification of WT, validating the feasibility of the model for clinical application. The segmentation results indicate that our model achieved a Dice of 0.95, 0.91, and 0.88 for the WT-blastema, WT-epithelium, and WT-stroma, respectively. The proposed model provides an automated solution for the histopathological classification of WT to assist pathologists in clinical diagnosis.
No takes yet. Share an insight, caveat, or question.
Wan et al. (2025) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: