Radiomics in Otolaryngology-Head and Neck Surgery: A Review.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
IMPORTANCE: Radiomics is a field that establishes associations between quantifiable imaging biomarkers and histopathological characteristics or clinical outcomes. Radiomics holds particular promise in otolaryngology given anatomic intricacies, diverse pathologies, and many commonplace imaging modalities. Radiomics applications span diagnostic classifiers, long-term prognosticators, and predictors of treatment response. The objective of this Review was to establish methodological frameworks, identify common limitations, and evaluate the current landscape of radiomics within otolaryngology. OBSERVATIONS: Radiomics applications span the breadth of otolaryngology, with most focused on neoplasms of the upper airway, larynx, sinonasal passages, and skull base. Head and neck cancer applications include classifiers of clinically occult pathologic features (such as extranodal extension or nodal metastases) that can guide treatment options. Prognostic radiomics can reliably model recurrence and survival outcomes, with hybrid clinical radiomics models achieving superior performance compared with single-modality models. Treatment prediction through approaches like dosiomics (using radiotherapy dose distributions) and δ-radiomics (sequential imaging over time) have shown potential in improving the prediction of therapeutic response, tumor recurrence, and radiotherapy toxic effects. Beyond neoplastic classifiers, a growing body of work has sought to risk stratify or predict the evolution of rhinologic and otologic inflammatory conditions (eg, chronic rhinosinusitis, middle ear disease). Recently, there have also been radiomics applications in sleep and pediatrics. Despite these broad advances, radiomics models have several pitfalls, such as variable imaging protocols, resource-intensive manual segmentation, limited cohort sizes, and a lack of external validation, all of which hinder clinical translation. CONCLUSIONS AND RELEVANCE: The results of this Review suggest that radiomics is a promising tool that can be integrated with clinical and pathologic data to enhance diagnosis, optimize prognostication, and personalize treatment in otolaryngology. Standardization of imaging protocols, rigorous validation in multi-institutional cohorts, and integration with clinical workflows remain critical prerequisites for clinical application. With continued refinement and integration, radiomics applications may help streamline clinical workflows and guide treatment planning.