Artificial intelligence-based CT-free quantitative thyroid SPECT for thyrotoxicosis: study protocol of a multicentre,...

Area of Science:

• Nuclear medicine
• Radiology
• Artificial intelligence

Background:

• Technetium thyroid uptake (TcTU) using SPECT/CT aids in differentiating Graves' disease from destructive thyroiditis.
• Artificial intelligence (AI) offers potential to reduce CT radiation exposure in SPECT/CT by enabling CT-free imaging for attenuation correction and thyroid segmentation.

Purpose of the Study:

• To compare the diagnostic accuracy of CT-free SPECT versus SPECT/CT for thyrotoxicosis differential diagnosis.
• To evaluate the non-inferiority of AI-based CT-free SPECT in diagnosing Graves' disease.

Main Methods:

• A prospective, single-blind, multicentre, non-inferiority clinical trial in the Republic of Korea.
• Adult thyrotoxicosis patients (n≈160) underwent both SPECT/CT and AI-based CT-free SPECT using Tc-99m pertechnetate.
• TcTU was calculated from both imaging methods; primary endpoint is diagnostic accuracy for Graves' disease.

Main Results:

• The study is ongoing, with a target of 152 completed datasets.
• The primary endpoint assesses if the accuracy of CT-free SPECT is non-inferior to SPECT/CT for Graves' disease diagnosis (lower confidence limit > -0.1).
• Secondary endpoints include diagnostic accuracy for destructive thyroiditis and prediction of antithyroid drug needs.

Conclusions:

• The study will determine if AI-enabled CT-free SPECT can accurately diagnose thyrotoxicosis subtypes, potentially reducing radiation dose.
• Successful implementation could make SPECT/CT imaging safer and more accessible for patients with Graves' disease and other thyrotoxic conditions.