PREOPERATIVE ASSESSMENT OF RECURRENT HEAD AND NECK CANCER RESECTABILITY: RADIOMICS, ARTIFICIAL INTELLIGENCE, AND MOLECULAR BIOMARKERS

Abstract

Introduction. Recurrent head and neck squamous cell carcinoma is difficult to assess after chemoradiotherapy because fibrosis, edema, and distortion of anatomical landmarks may obscure viable tumor. Aim. To summarize the roles of computed tomography, magnetic resonance imaging, positron emission tomography, radiomics, artificial intelligence, and molecular biomarkers in preoperative assessment of resectability and complication risk. Materials and methods. A narrative synthesis of 39 verified and clinically relevant publications was performed. Sources were grouped by imaging modality, radiomic modeling, molecular phenotype, and validation requirements. Results. Conventional imaging remains essential for defining tumor extent, but specificity decreases in the presence of post treatment change. Radiomic features may capture tumor heterogeneity, and radiogenomic studies have reported associations with human papillomavirus status and molecular subtypes. Clinical translation remains limited by heterogeneous imaging protocols, overfitting, inadequate external validation, and the lack of prospective evidence that model-assisted decisions improve surgical outcomes. Conclusion. Standardized multiparametric imaging should be integrated with clinical variables and validated molecular biomarkers. Until prospectively evaluated, these models should support rather than replace multidisciplinary decisions regarding salvage surgery.