Based on ADC histogram for distinguishing benign and malignant thyroid nodules: A small FNA-indeterminate cohort exploratory investigation

Motivation: In clinical practice, there are still some patients with uncertain FNA results and difficult to decide on the next treatment plan Goal(s): Use routine MRI sequences and simple model to supplement ultrasound TIRADS standards results assist clinicians in preoperative decision-making Approach: The ADC images histogram features combined with MRI morphological features to construct a regression model to determine the benign and malignant thyroid nodules. The results were compared with TIRADS standards Results: The ADC-morphological feature model had great diagnostic performance in distinguishing Identifying the benign or malignant thyroid nodules (AUC 0.91).It shown higher specificity (84.2%) than that of the TIRADS (61.90%) Impact: The prediction model based on ADC and morphological features shows a higher specificity than ultrasound TIRADS standards. ADC parameter histogram features can evaluate the benign and malignant nodules, especially nodules with uncertain FNA, assist clinicians in reducing unnecessary biopsy rates.