Clinical values of of gemstone spectral CT in diagnosing thyroid disease

Abstract

OBJECTIVE: To quantitatively evaluate the diagnostic value of gemstone spectral CT in thyroid disease. PATIENTS and METHODS: A total of 123 patients with thyroid diseases were enrolled in the retrospective study. All the patients underwent spectral CT scan, and 39 of them underwent dual-phase enhanced scan. Iodine concentration (IC) and normalized IC (NIC) were compared between benign and malignant nodules. The optimal threshold to predict malignancy was obtained by receiver operating characteristic curve (ROC). Multivariate ROC analysis was performed to evaluate the efficacy of combining the IC (NIC) and conventional morphological characteristics. RESULTS: Ten diffuse diseases and 113 nodular diseases were confirmed by clinical laboratory examination and histopathology. In total, 122 nodules (87 benign and 35 malignant) were detected, 41 nodules in enhanced cases. The IC and NICVP of malignant nodules were significantly lower than benign nodules in noncontrast and venous phase (VP) (P = 0.000). The optimal IC and NICVP threshold was 0.68 mg/mL and 0.38, achieving 74.3% and 76.9% sensitivity, 66.7% and 80.0% specificity, respectively. For multivariate analysis, the area under ROC (AUC) was 0.88 and 0.94 in noncontrast and VP, respectively. Sensitivities of 77.1% and 84.6%, and specificities of 84.5% and 88.0% were achieved. CONCLUSION: Gemstone spectral CT can generally evaluate the abnormal iodine content in thyroid disease. Lower iodine can help differentiating malignant from benign nodules. A comprehensive analysis combining spectrum parameters and morphological characteristics is suggested in differential diagnosis.

Dual‐energy CT iodine quantification for characterizing focal thyroid lesions

“…To our knowledge, a few studies have addressed the relationship between the quantitative parameters derived from DECT and thyroid disease, for example, IC, effective atomic numbers (Z eff ), or spectral curve slopes of thyroid nodules. [23][24][25][26] Li et al first reported significant differences in IC, Z eff , or spectral curve slopes between malignant and benign thyroid nodules using non-enhanced CT images with 97 surgically removed thyroid nodule specimens (all P < .001). 23 Later, the same group suggested practical parameters regarding the regional characteristics of thyroid nodules, such as a threshold IC value for discriminating intranodular hemorrhage from the surrounding solid portion, which was not apparent by single-energy CT. 24 According to another study, both IC and normalized IC values were significantly lower in malignant nodules than benign nodules (P < .001); therefore, the thresholds for discrimination were determined to be about 0.68 mg/mL for IC and 0.38 mg/mL for normalized IC, with a sensitivity of 74.3%/76.9% and a specificity of 66.7%/80.0%, respectively.…”

“…23 Later, the same group suggested practical parameters regarding the regional characteristics of thyroid nodules, such as a threshold IC value for discriminating intranodular hemorrhage from the surrounding solid portion, which was not apparent by single-energy CT. 24 According to another study, both IC and normalized IC values were significantly lower in malignant nodules than benign nodules (P < .001); therefore, the thresholds for discrimination were determined to be about 0.68 mg/mL for IC and 0.38 mg/mL for normalized IC, with a sensitivity of 74.3%/76.9% and a specificity of 66.7%/80.0%, respectively. 25 However, their diagnostic performances of IC varied substantially depending on the degree of contrast enhancement; therefore, the AUC of contrast-enhanced CT with a 33-to 36-second scan delay was greater than that of the non-enhanced images (0.88 vs 0.71), whereas normalized ICs obtained from relatively early scans, for example, 20-to 23-second scan delay, showed no difference between the 2 groups (P = 0.65). In this regard, scan delay would be important in determining the diagnostic performance of IC; moreover, sufficient tumor conspicuity must be required for accurate ROI drawing on the contrast-enhanced CT image.…”

“…To our knowledge, a few studies have addressed the relationship between the quantitative parameters derived from DECT and thyroid disease, for example, IC, effective atomic numbers (Z eff ), or spectral curve slopes of thyroid nodules . Li et al first reported significant differences in IC, Z eff , or spectral curve slopes between malignant and benign thyroid nodules using non‐enhanced CT images with 97 surgically removed thyroid nodule specimens (all P < .001) .…”

“…[19][20][21][22] Until now, a few studies have evaluated IC of thyroid nodules using the DECT data sets to find the malignancy or intranodular hemorrhage; however, they decided the individual CT images for quantification without the detailed topographic knowledge based on ultrasound, which is widely considered as reference imaging modality to assess thyroid nodule. [23][24][25][26] In order to reduce the potential mismatch between the DECT images and other diagnostic methods, representative images of thyroid nodule have to be selected meticulously. Therefore, we carefully identified the focal thyroid lesions of which size, shape, and position coincided with thyroid nodules that were previously diagnosed based on their relevant sonographic appearances and cytopathological results.…”

“…In the head and neck region, several studies showed iodine concentration (IC) could be useful in CT imaging of cervical lymph nodes or patients with underlying neck malignancy . Until now, a few studies have evaluated IC of thyroid nodules using the DECT data sets to find the malignancy or intranodular hemorrhage; however, they decided the individual CT images for quantification without the detailed topographic knowledge based on ultrasound, which is widely considered as reference imaging modality to assess thyroid nodule …”