鼻咽癌
肿瘤科
淋巴结
预测模型
医学
生存分析
内科学
无线电技术
危险分层
总体生存率
计算生物学
栖息地
原发性肿瘤
病理
生物
序列(生物学)
空间异质性
多元分析
淋巴结转移
癌
节点(物理)
形态分析
通路分析
空间关系
预测建模
作者
Zijun Huang,Yu Li,Jia Kou,Shanqi Bao,Sun Y,Li Li
出处
期刊:Bioengineering
[Multidisciplinary Digital Publishing Institute]
日期:2026-04-29
卷期号:13 (5): 521-521
标识
DOI:10.3390/bioengineering13050521
摘要
Habitat analysis enables spatial characterization of intratumoral heterogeneity; however, its application in nasopharyngeal carcinoma (NPC), particularly regarding metastatic lymph node (MLN), remains limited. This study aims to systematically compare the prognostic performance of various models using different sequences and spatial region combinations for predicting overall survival in NPC. The study retrospectively included 725 NPC patients (543 training, 182 testing). Habitat analysis was conducted based on T1, T1C, and T2 sequences in three regional strategies: primary gross tumor volume (GTVp), metastatic lymph nodes (MLNs), and the combined region of GTVp-MLN. The tumor area was divided into six subregions, and a multi-region spatial interaction (MSI) matrix was constructed to extract MSI features. On this basis, a radiomics model (R Model) and a clinical-radiomics model (CR Model) were established, and the model performance was evaluated using C-index and Kaplan-Meier survival analysis. The results show that the combined GTVp-MLN model based on the T1 sequence achieved the best overall predictive performance (R Model: C-index = 0.693; CR Model: C-index = 0.722). Significant survival differences were observed between the high- and low-risk groups. These findings suggest that habitat analysis incorporating the combined GTVp-MLN region may improve prognostic prediction and risk stratification in patients with NPC.
科研通智能强力驱动
Strongly Powered by AbleSci AI