thyroid-cancer

AI diagnostic system matches expert thyroid ultrasound performance—and could close the gap in community hospitals.

AI diagnostic system for thyroid nodules C-TIRADS ≥4A achieved excellent agreement with expert teleultrasound (kappa=0.80), vs. poor agreement between community institutions and teleultrasound (kappa=0.20). The AI achieved sensitivity of 97.1% and specificity of 100% at ≥C-TIRADS 4A threshold; AUC…

Dan Manning

Dan Manning

— 1 min read
Share
AI diagnostic system matches expert thyroid ultrasound performance—and could close the gap in community hospitals.

AI diagnostic system matches expert thyroid ultrasound performance—and could close the gap in community hospitals.

AI diagnostic system for thyroid nodules C-TIRADS ≥4A achieved excellent agreement with expert teleultrasound (kappa=0.80), vs. poor agreement between community institutions and teleultrasound (kappa=0.20). The AI achieved sensitivity of 97.1% and specificity of 100% at ≥C-TIRADS 4A threshold; AUC of 0.92—comparable to teleultrasound experts.

Key Findings

  • AI vs. expert teleultrasound: strong agreement (kappa 0.80)
  • Community institutions vs. teleultrasound: poor agreement (kappa 0.20)
  • AI AUC 0.92 at ≥C-TIRADS 4A—comparable to teleultrasound
  • AI sensitivity 97.1%, specificity 100% at ≥C-TIRADS 4A threshold
  • System could close diagnostic gap in community hospitals for thyroid cancer screening

Implications

AI-assisted teleultrasound could improve diagnostic consistency for thyroid nodules in resource-limited community settings, reducing the gap between community and expert-level care.

Caveats

Retrospective (n=587 nodules); abstract-only. Performance validated at single center. Generalizability to other C-TIRADS systems needs assessment.

Source: Frontiers in endocrinology — 2026-01-01

Read more