AI Can Now Read Thyroid Scans to Avoid Unnecessary Surgery
This could save you from surgery and reduce healthcare costs.
When a thyroid nodule (a lump in the thyroid gland) is found, doctors often do a biopsy. But sometimes the biopsy results are unclear—not clearly benign, not clearly cancerous. These are called indeterminate nodules.
Currently, doctors may use expensive molecular tests to decide whether surgery is needed. But these tests cost thousands of dollars and aren't available everywhere. Now, researchers have tested an AI model that looks at standard ultrasound images to help make that decision.
In their study, the AI was better than molecular tests at correctly identifying nodules that were not cancer, meaning it could help some people avoid unnecessary surgery. However, the AI also missed more cancers than the molecular tests did, so it's not perfect. The AI was especially good at flagging which molecular-test-positive nodules were actually low-risk, which could spare some people from surgery. The study is small, so more research is needed before this AI becomes routine. But it's a promising step toward cheaper, more accessible thyroid cancer diagnosis.
- An AI model analyzing ultrasound images was better than molecular tests at ruling out cancer in indeterminate thyroid nodules, potentially avoiding unnecessary surgeries.
- The AI missed more cancers than molecular tests, so it's not a replacement yet.
- If validated, this could make diagnosis cheaper and more accessible, especially where molecular tests aren't available.
Why It Matters
This could reduce unnecessary surgeries and healthcare costs for thyroid patients.