JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Diagnostic accuracy of radiomics and artificial intelligence models in diagnosing lymph node metastasis in head and neck cancers: a systematic review and meta-analysis

  • 0School of Medicine, Tehran University of Medical Sciences, Tehran, Iran.
Neuroradiology +

|

November 11, 2024

|

November 11, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Artificial intelligence (AI) models show high accuracy in detecting lymph node metastasis (LNM) in head and neck cancers, especially when using PET/CT imaging. Further validation is needed for clinical use.

Area Of Science

  • Oncology
  • Medical Imaging
  • Artificial Intelligence

Background

  • Head and neck cancers are a significant global health concern, with lymph node metastasis (LNM) being a key factor affecting patient survival.
  • Conventional imaging techniques often struggle with accurate LNM detection, necessitating advanced diagnostic approaches.

Purpose Of The Study

  • To evaluate the diagnostic accuracy of Artificial Intelligence (AI) models for detecting lymph node metastasis (LNM) in head and neck cancers.
  • To synthesize existing evidence through a meta-analysis of AI-based diagnostic performance.

Main Methods

  • A systematic literature search was conducted across four databases for studies on AI models detecting LNM in head and neck cancers.
  • The METRICS tool assessed study quality, and a bivariate model in R was used for meta-analysis.

Main Results

  • The meta-analysis included 23 articles, focusing on internal validation sets.
  • Pooled AUCs were high: 91% for CT-radiomics, 84% for MRI-radiomics, and 92% for PET/CT-radiomics.
  • Deep learning and hand-crafted radiomics models, as well as models using lymph node or primary tumor features, demonstrated comparable high performance (AUCs 89-92%).

Conclusions

  • AI models, particularly radiomics and deep learning, show significant promise for diagnosing LNM in head and neck cancers, with PET/CT-based models being especially effective.
  • Future research should emphasize multicenter studies with external validation to improve clinical applicability.

Keywords: