Systematic Review of AI-Driven Applications for Screening Nasopharyngeal Carcinoma Using MRI.
Lun M Wong, Ho Sang Leung, Zuyao Yang et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review evaluated the diagnostic performance of AI applications for screening nasopharyngeal carcinoma (NPC) using MRI, analyzing 38 studies involving 23,398 patients. The findings indicated a pooled Dice similarity score of 0.80 for lesion localization and a sensitivity of 97.1% and specificity of 87.8% for NPC discrimination. Only seven studies were fully applicable to screening.
Key findings
- Pooled Dice similarity score for lesion localization was 0.80 (CI: 0.78-0.82).
- Sensitivity for NPC discrimination was 97.1% (CI: 88.0%-99.3%).
- Specificity for NPC discrimination was 87.8% (CI: 78.9%-93.2%).
- Intravenous contrast did not significantly affect AI performance (p > 0.05).
Why it matters
The high sensitivity of AI-driven MRI applications for NPC detection suggests potential for improving early diagnosis, particularly in settings with limited access to experienced specialists. Understanding the applicability of these technologies in screening can enhance patient outcomes.
What to keep in mind
Only seven studies were fully applicable to screening, indicating a need for further research in more representative cohorts.