Malignancy classification of thyroid incidentalomas using 18 F-fluorodeoxy- d -glucose PET/computed...

Area of Science:

• Nuclear Medicine
• Radiology
• Oncology

Background:

• Thyroid incidentalomas (TIs) are common findings on 18F-FDG PET/CT scans.
• Characterizing these lesions is crucial for patient management in oncological settings.

Purpose of the Study:

• To evaluate the utility of noninvasive PET/CT-derived radiomic features in distinguishing benign from malignant thyroid incidentalomas.
• To compare the performance of radiomics with the ultrasound TIRADS classification.

Main Methods:

• 46 patients with PET/CT TIs underwent ultrasound and surgery.
• Radiomic features were extracted from PET and CT images.
• Histopathology served as the reference standard; analyses included univariate, multivariate, and comparative assessments against TIRADS.

Main Results:

• Malignancy was found in 78% of TIs.
• A combined radiomic feature (run length nonuniformity) achieved an AUC of 0.88, comparable to TIRADS (AUC: 0.84).
• CT-derived radiomics demonstrated superior predictive performance (AUC: 0.86) compared to TIRADS (AUC: 0.80) in multivariate analysis.

Conclusions:

• 18F-FDG PET/CT-derived radiomics show significant potential for differentiating benign and malignant thyroid lesions.
• Larger cohorts and advanced methods like deep learning are recommended for further validation.