Scholarlink

<p>Radiogenomics links quantitative features from routine CT and PET/CT with tumor genomics to non-invasively profile non-small cell lung cancer (NSCLC). This review synthesizes the current workflow—from image acquisition and segmentation to feature extraction and modeling—and emphasizes clinical use cases: triage when tissue is limited, risk stratification, therapy selection (including immunotherapy), and longitudinal monitoring. Despite promising results, clinical translation is constrained by non-standardized imaging/feature pipelines, limited multi-center validation and calibration, vulnerability to data leakage, and challenges in interpretability, uncertainty handling, and software and model governance. We advocate treating genomic burden—tumor mutational burden (TMB), intratumor heterogeneity (ITH), and copy-number alterations (CNA)—as first-class endpoints and covariates, and outline a pattern-aware framework for EGFR T790M to inform surveillance and treatment sequencing. We also provide a practical reporting checklist and a pitfall-to-remedy table to support reproducible, multi-center studies and regulatory-grade documentation. Radiogenomics is best viewed as a complement to biopsy rather than a replacement. Real-world impact now depends on harmonized protocols, leakage-free external validation, explainable and uncertainty-aware models, and integration with multi-omics decision support to deliver reliable, patient-centered lung cancer care.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Radiogenomics: transforming lung cancer care through non-invasive imaging and genomic integration

  • 期刊论文

  • 发表日期：

    2025年11月14日

查看全文
  • Faridoddin Shariaty,
  • Vitalii Pavlov

摘要

Radiogenomics links quantitative features from routine CT and PET/CT with tumor genomics to non-invasively profile non-small cell lung cancer (NSCLC). This review synthesizes the current workflow—from image acquisition and segmentation to feature extraction and modeling—and emphasizes clinical use cases: triage when tissue is limited, risk stratification, therapy selection (including immunotherapy), and longitudinal monitoring. Despite promising results, clinical translation is constrained by non-standardized imaging/feature pipelines, limited multi-center validation and calibration, vulnerability to data leakage, and challenges in interpretability, uncertainty handling, and software and model governance. We advocate treating genomic burden—tumor mutational burden (TMB), intratumor heterogeneity (ITH), and copy-number alterations (CNA)—as first-class endpoints and covariates, and outline a pattern-aware framework for EGFR T790M to inform surveillance and treatment sequencing. We also provide a practical reporting checklist and a pitfall-to-remedy table to support reproducible, multi-center studies and regulatory-grade documentation. Radiogenomics is best viewed as a complement to biopsy rather than a replacement. Real-world impact now depends on harmonized protocols, leakage-free external validation, explainable and uncertainty-aware models, and integration with multi-omics decision support to deliver reliable, patient-centered lung cancer care.