Optimized dual-source dual-energy computed tomography nomogram model integrating background normalization improves solid thyroid nodules diagnosis

Introduction

The workload of diagnostic imaging specialists in identifying and managing malignant thyroid nodules is rapidly increasing due to the rising incidence of thyroid nodules, which are detected in approximately 65% of the general population (1,2). Given the differences in composition, treatment, and post-diagnosis outcomes between benign and malignant nodules, early and accurate diagnosis is essential for effective clinical management and optimal prognosis. Ultrasound remains the primary imaging modality for thyroid nodules, offering a noninvasive, convenient, and cost-effective approach. However, its reliability in assessing deep neck anatomy, cervical lymph node status, and operator dependence remains debated (3,4). Using ultrasonography alone to identify benign and malignant nodules, as well as grading and staging malignant nodules, is unstable in clinical practice (5,6). While fine-needle aspiration biopsy (FNAB) is the diagnostic gold standard, there are limitations related to sampling and invasiveness (7-9). Magnetic resonance imaging (MRI) offers high resolution and multimodal capabilities but is time-consuming and prone to motion artifacts (10,11). Therefore, one of the current unavoidable clinical issues is the need for a non-invasive and accurate method to differentiate between multiple thyroid nodules preoperatively while ensuring reproducibility and providing reliable anatomical and lymph node information for treatment planning. Enhanced computed tomography (CT) is recommended as a helpful supplementary examination for some patients (1). However, accurately distinguishing between benign and malignant thyroid nodules using conventional CT remains challenging for some diagnosticians in real-world clinical practice (12). Dual-energy CT (DECT), an emerging imaging technology, improves image quality without increasing radiation exposure and quantifies tissue biological properties through iodine maps and energy spectral curves (12,13). Some studies have shown DECT’s efficacy in differentiating benign from malignant thyroid nodules and assessing regional lymph node metastasis (14-18). Nevertheless, existing studies have largely focused on the morphological features of nodules and selected quantitative parameters, often neglecting the influence of the surrounding thyroid tissue.

Our study aimed to derive additional quantitative metrics with diagnostic relevance by standardizing measurements against background thyroid tissue. We hypothesized that DECT-derived quantitative parameters related to surrounding tissues can assist in distinguishing between benign and malignant nodules. To test this hypothesis, we developed a nomogram model integrating DECT-derived quantitative parameters with radiological features and internally validated the predictive model’s effectiveness. The objective is to provide radiologists and clinicians with an intuitive and efficient diagnostic tool to optimize clinical diagnosis and treatment strategies. We present this article in accordance with the TRIPOD reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2025-421/rc).

Methods

Patients

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The First People’s Hospital of Lianyungang (approval No. KY-20220506001-01). Informed consent was waived in this retrospective study. Data were consecutively collected from patients who underwent neck DECT (SOMATOM Drive, Siemens Healthineers, Forchheim, Germany) scanning at The First People’s Hospital of Lianyungang between December 2017 and March 2023, following predefined inclusion and exclusion criteria. Inclusion criteria included: (I) dual-phase enhancement of DECT and neck CT scanning within one week prior to the procedure; (II) pathological confirmation of benign or malignant thyroid nodules within two weeks post-procedure; (III) patients aged 18 years or older; and (IV) selection of the largest nodule when multiple nodules were present in a single thyroid lobule, with a minimum diameter of ≥1 cm. Exclusion criteria were: (I) a history of additional malignant tumors (n=36) or prior thyroid interventions; (II) artifacts or noise in low-quality DECT images (n=19); and (III) quantitative assessments affected by predominant calcification or cystic changes (n=31). Clinical data, and surgical pathology outcomes, were extracted from patient medical records. The dataset was further stratified into a training set (n=201, 70%) and a validation set (n=87, 30%) in a 7:3 ratio. Figure 1 illustrates the screening flowchart.

Figure 1 Flow chart of the study. CT, computed tomography; DECT, dual-energy computed tomography.

DECT image acquisition

The study’s participants received both standard neck CT plain scanning and arteriovenous dual-phase DECT scanning from the base of the skull to the aortic arch. Acquisition protocol: Automatic tube current modulation (CARE Dose 4D) was activated, linear fusion factor was set to 0.6, weighted fusion maps were automatically generated after scanning, and bulb tube voltages A and B were set to 90 kV and Sn150 kV, respectively. A helical pitch of 0.9, a matrix of 256×256, rotation period of 0.5 s/turn, detector collimation layer thickness of 128×0.6 mm, and layer thickness 5 mm. Monoenergetic image reconstruction convolution kernel (Kernel): Q30f medium smooth, iterative reconstruction procedure for ADMIRE, reconstructed layer thickness 1.0 mm. A follow-up injection of 30 mL of saline at a flow rate of 2.5–3.5 ml/s was performed after a non-ionic contrast agent (1.5 m L/kg ioversol, 320 mg/mL iodine, Jiangsu Hengrui Pharmaceuticals, Lianyungang, China) was administered through the elbow vein using a high-pressure syringe regimen. Following a standard CT scan, the patient had arterial and venous dual-phase dual-energy scanning. The Smart Trigger Scanning system was applied with a delayed time; 25–30 seconds after the venous phase scanning, scanning was automatically initiated when the aortic arch’s CT value reached 100 Hounsfield unit (HU).

Radiological features and DECT quantitative data analysis

A team of diagnostic radiologists, consisting of two attending physicians [observer 1 (Y.X.) and observer 2 (Y.F.)] and a chief physician (Y.G.) with over ten years of experience in CT diagnostics, conducted the imaging characterization of all nodules. The two attending physicians independently evaluated the presence or absence of five imaging signs: enhanced blurring, punctate calcification, irregular shape, low density on plain CT scan, and thyroid edge interruption. They only knew the location and diameter of the nodules, without knowledge of the surgical pathology. Irregular shape indicated the nodule was non-circular or non-oval. Low density defined the nodule having a density lower than the background thyroid tissue on plain CT scan. Punctate calcification was classified as calcifications that were less than 2 mm in diameter. Thyroid edge interruption, sometimes known as the “Bitten-cookie” sign (19), is a localized defect in the thyroid gland’s rim that is replaced on plain or enhanced CT scans by a relatively low-density tumor. Enhanced blurring means the boundary between the tumor and the background thyroid tissue was blurred after enhancement (20). Five radiological signs were assessed on both plain and enhanced CT images. In cases of disagreement between the two attending physicians regarding image interpretation, the chief physician provided the final consensus.

All DECT quantitative parameters were post-processed using the Synago.via workstation. The arterial and venous phase images were independently measured twice by the same two radiologists [observer1 (Y.X.) and observer 2 (Y.F.)], with the average of these measurements used for subsequent analysis. The observer manually placed the region of interest (ROI) on the maximum solid component of the nodule, avoiding surrounding fat, blood vessels, internal calcifications, cystic degeneration, and necrosis. The location, size, and shape of the ROI remained consistent when measuring the quantitative parameters of the same nodule. The virtual nonenhanced mode (also known as the “Liver VNC” mode) and the Rho/Z mode were used to measure the IC and Zeff values of the lesions in the arterial and venous phases, respectively. The IC and Zeff values of the background thyroid parenchyma and the common carotid artery were also evaluated at the same level. The IC and Zeff values of the lesions were normalized to the common carotid artery and the background thyroid tissue on the same level to reduce the influence caused by individual differences in hemodynamics and hormone levels. The following are the formulas: NIC_CCA = IC of nodule/IC of the common carotid artery, NIC_P = IC of nodule/IC of the thyroid parenchyma, NZeff_CCA = Zeff of nodule/Zeff of the common carotid artery, NZeff_P = Zeff of nodule/Zeff of the thyroid parenchyma.

Furthermore, single-energy CT values and energy spectra of the nodes were acquired, together with pictures of the arterial and venous phases’ energy spectra (40 to 190 keV at 10 keV intervals) produced in the “Mono+” mode. Energy spectrum’s slope: λHU = (CT value at 40 keV − CT value at 100 keV)/(100 − 40). The following is a shorthand for the DECT parameters: CCA = common carotid artery, P = thyroid parenchyma, IAP = in the arterial phase, IVP = in the venous phase, IC = iodine concentration, NIC = normalized iodine concentration, Zeff = effective atomic number, NZeff = normalized effective atomic number, λHU = slope of the spectral HU curve.

Development and assessment of the nomogram models

Initially, univariate logistic regression analysis was performed on the qualitative imaging features of thyroid nodules and DECT quantitative parameters to identify potential malignancy predictors (P<0.05). Subsequently, the least absolute shrinkage and selection operator (LASSO) logistic regression algorithm was applied to select significant features with non-zero coefficients. Ten-fold cross-validation determined the optimal parameter configuration, using the lambda value at one standard error (1 SE) to filter non-zero coefficient variables. For the selected variables, a multivariate logistic regression analysis (stepwise, bidirectional) was conducted to identify independent predictors for malignancy assessment (P<0.05). Regression coefficients were used to create nomograms that combined DECT quantitative and radiological features, along with separate models based on independent DECT and radiological variables. All three models were validated in an independent cohort. To assess their accuracy, the nomograms underwent 1000 bootstrap sampling validation, generating calibration curves. Model fit was evaluated using the Hosmer-Lemeshow test, while receiver operating characteristic (ROC) curve analysis provided metrics such as area under the curve (AUC), sensitivity, and specificity for discriminative performance. The DeLong test was applied for AUC comparison among the models. Additionally, clinical decision curve analysis (DCA) was developed to evaluate the models’ clinical utility, quantifying net benefits at specified threshold probabilities. Ultimately, the performance of all three models was confirmed in the validation cohort.

Statistical analysis

Data processing and statistical analysis were performed using R software (version 4.3.2) and SPSS 25.0 (version 25.0, IBM Corp., Armonk, NY, USA). The “glmnet” package was utilized for LASSO analysis, while the “rms” package was employed for nomogram construction and calibration curve generation. ROC curves were plotted using the “pROC” and “ggplot2” packages. Clinical DCA was conducted using the “ggDCA” package. The normality of continuous variables was assessed using the Kolmogorov-Smirnov test, and the homogeneity of variances was evaluated using Levene’s test. As most continuous variables were not normally distributed, group differences were analyzed using the non-parametric Mann-Whitney U test. Continuous data are presented as mean (standard deviation). Chi-squared tests were used for comparisons of categorical data. A significance level of P<0.05 (two-tailed) was adopted for determining statistical significance. The cohort was randomly partitioned via the R sample function into training (70%) and validation (30%) subsets, with the former utilized for training and the latter for performance evaluation. The intraclass correlation coefficient (ICC) was used to evaluate the reliability of quantitative parameters values in both intra-observer and inter-observer assessments, with values greater than 0.80 indicating good agreement.

Results

Clinical characteristics

A total of 288 nodules from 263 eligible patients were analyzed, comprising 209 females and 54 males, aged 20 to 76 years, with a median age of 51 years and a mean age of 48.74±12.66 years. Histopathological findings revealed 158 benign nodules (132 adenomas, 17 nodular goiters, and 9 nodular goiters with adenomatous hyperplasia) and 130 malignant nodules (123 papillary carcinomas and 7 follicular carcinomas), with 48 patients also diagnosed with Hashimoto’s thyroiditis (HT). There were no adverse outcomes from this study.

Comparison of DECT parameters and the typical radiological features between benign and malignant nodules

Table 1 provides a detailed summary of the radiological characteristics and quantitative DECT parameters for the training and validation datasets. Fifteen candidate parameters were identified, including three qualitative imaging features and 12 quantitative DECT parameters, as revealed by univariate analysis of the training dataset (P<0.05) (Table 2). All quantitative parameters values demonstrated good reproducibility for both intraobserver and interobserver assessments (Table S1). Additionally, the quantitative parameters of the arterial and venous phases of malignant nodules: IC, NIC_CCA, NIC_P, Zeff, NZeff_P, and λHU, were significantly lower than those of the benign nodule group (all P<0.05) (Table S2).

Using the LASSO logistic regression algorithm, we identified eight variables with non-zero coefficients significant features as potential predictors (Figure S1). Subsequent multivariate logistic regression analysis revealed five statistically independent predictors differentiating benign from malignant solid nodules (all P<0.05): thyroid edge interruption [odds ratio (OR) =3.791, 95% confidence interval (CI): 1.599–9.491, P=0.003], enhanced blurring (OR =3.247, 95% CI: 1.373–7.937, P=0.008), IC_IAP (OR =0.286, 95% CI: 0.140–0.517, P<0.001), NIC_P_IAP (OR =0.133, 95% CI: 0.033–0.471, P= 0.003), and Zeff_IVP (OR =0.137, 95% CI: 0.050–0.333, P<0.001) (Table 3).

Nomogram construction and performance

Based on the aforementioned five independent risk factors, a radiological-DECT nomogram was created to predict the likelihood of a malignant nodule occurrence (Figure 2). A higher total nodule score indicates a higher risk of malignant nodule occurrence. Figure 3 shows an illustration of representative DECT images of nodules by use of nomogram prediction. The prediction model achieved the highest AUC values among the three models, with an AUC of 0.932 (95% CI: 0.900–0.965), a sensitivity of 0.870 (95% CI: 0.801–0.938), and a specificity of 0.862 (95% CI: 0.798–0.927) in the training set, along with corresponding positive predictive value and negative predictive value were shown in Table 4. Figure 4 illustrates the ROC curves and AUC values for these models.

Figure 2 A DECT-radiological diagnostic nomogram. The nomogram was plotted combining two radiological features and three DECT quantitative parameters in the training cohort. The method for calculating the risk of malignant thyroid nodules was as follows: first, points for each predictor are assigned by corresponding values from the “Points” axis. Second, the “Total points” is obtained by summing up points of all predictors. Third, a vertical line is drawn down the “Total points” to determine the risk of malignant thyroid nodules. 0= no; 1= yes. DECT, dual-energy computed tomography; IC_IAP, iodine concentration in the arterial phase; NIC_P_IAP, normalized iodine concentration relative to the thyroid parenchyma in the arterial phase; Zeff_IVP, effective atomic number in the venous phase.

Figure 3 Representative examples of correctly classified nodules by the nomogram. (A-C) True-negative case: thyroid adenoma. Patient: 60-year-old female. Location: right lobe (green arrow). Parameters: IC_IAP_nodule is 6.0 mg/mL, NIC_P_IAP_nodule is 1.25, Zeff_IVP_nodule is 9.65. Nomogram output: malignancy probability =0.015. (D-F) True-positive case: papillary thyroid carcinoma. Patient: 34-year-old female. Location: left lobe (red arrow). Parameters: IC_IAP_nodule is 1.6 mg/mL, NIC_P_IAP_nodule is 0.609, Zeff_IVP_nodule is 8.51. Nomogram output: malignancy probability =0.986. DECT, dual-energy computed tomography; IC_IAP, iodine concentration in the arterial phase; NIC_P_IAP, normalized iodine concentration relative to the thyroid parenchyma in the arterial phase; Zeff_IVP, effective atomic number in the venous phase.

Figure 4 Diagnostic efficacy of combined model versus single model for differentiating common thyroid nodules. (A) ROC curves of the radiological model, quantitative DECT model, and DECT-radiological nomogram in the training cohort; (B) ROC curves of three models in the validation cohorts. AUC, area under the curve; DECT, dual-energy computed tomography; ROC, receiver operating characteristic.

The DeLong test indicated that the quantitative model and the DECT-radiological nomogram outperformed the radiological model (P<0.001), though no significant difference was found between the first two (P=0.16) (Table S3). The Hosmer-Lemeshow test indicated no significant lack of fit in the training set (χ²=5.58, P=0.69) and validation set (χ²=4.45, P=0.81). Calibration curves further supported model accuracy, both the training cohort and the validation cohort closely approximating the ideal line. The radiological-DECT nomogram demonstrated concordance between predicted probabilities and actual pathological outcomes, as visualized in Figure 5. Additionally, DCA revealed that both cohorts significantly benefited from the predictive model (Figure 6).

Figure 5 Calibration curves for the DECT-radiological nomogram. (A) Calibration curve derived from the training cohort. (B) Calibration curve derived from the validation cohort. The 45° straight line indicates the ideal performance of the DECT-radiological nomogram. A closer distance between two curves indicates higher accuracy. DECT, dual-energy computed tomography.

Figure 6 Decision curve analysis comparing the clinical net benefit of the 3 models. (A) Results in the training cohort. (B) Results in the validation cohort. The DECT-radiological nomogram and the DECT model demonstrated a higher overall net benefit across most threshold probabilities compared to the radiological model. DECT, dual-energy computed tomography.

Discussion

In our study, we developed and internally validated a nomogram that integrates two conventional CT qualitative features with three DECT quantitative parameters to differentiate between common classical benign and malignant thyroid nodules, and it showed good evaluation performance. This indicates that DECT parameters may augment the diagnostic accuracy of conventional CT imaging, offering supplementary diagnostic value for classifying high-risk nodules. Furthermore, our quantitative DECT model also showed superior diagnostic efficacy compared to qualitative radiographic characterization models, consistent with earlier findings (20,21). In both the training and validation cohorts, the nomogram model that incorporates both quantitative and qualitative multidimensional data demonstrates superior performance compared to single-feature models. This model is particularly well-suited for first-line diagnostic radiologists who encounter atypical radiological features of nodules, face challenges in qualitative ultrasound assessment, or possess limited experience in this domain. The calibration curve prediction results of the nomogram model are in good agreement with the actual observation results, which is a more reliable method to improve the diagnostic accuracy. It is crucial to remember, though, that varying scanning circumstances could lead to various tissue separations. To evaluate the clinical applicability of this technological approach, a comparative analysis must be conducted and modifications made as needed.

Prior research has indicated that quantitative parameters derived from DECT can be effectively utilized in the differential diagnosis of malignant and benign tumors (15,21,22). In the current study, we also identified that quantitative parameters extracted from DECT imaging, including IC, NIC_CCA, NIC_P, Zeff, NZeff_P, and λHU, could be used to differentiate between common benign and malignant thyroid nodules. Further analysis and ROC revealed that IC_IAP exhibited the highest diagnostic efficiency, with an AUC of 0.820 and a sensitivity of 0.981. Zeff_IVP demonstrated the best specificity at 0.908, albeit with lower sensitivity and AUC. NIC_P_IAP showed high accuracy, AUC, and specificity. Variables derived from IC primarily reflect local blood volume and are influenced by the quantity and distribution of blood vessels (23). Zeff is a quantitative measure that primarily represents a compound’s atomic number. Generally, materials containing elements with higher atomic numbers have a higher Zeff (24). The number and location of blood arteries have an impact on the derived variables connected to IC, which primarily represent local blood volume. Our study confirms that IC_IAP, Zeff_IVP and NIC_P_IAP can be used as independent predictors of differential diagnosis of common benign and malignant thyroid nodules. Recent research has revealed that IC and Zeff-derived variables correspond with the expression of certain immunohistochemical markers in tumors, including breast cancer and rectal cancer (25,26), and have high diagnostic value in the prediction and differentiation of metastatic lymph nodes in tumors (27,28), which also opened up the direction for our later research. However, unlike malignant lung nodules, breast cancer, rectal cancer, and other blood rich tumors, the quantitative parameters of malignant thyroid nodules are generally lower than those of benign nodules. This may be because the tumor cells of malignant thyroid nodules destroy thyroid follicular cells more extensively than those of benign nodules, which leads to decreased iodine uptake function of lesions. This effect may exceed the increase in quantitative parameters caused by tumor neoangiogenesis.

Notably, several retrospective investigations have reported a high coexistence rate of Hashimoto thyroiditis and thyroid carcinoma (29,30), of roughly 23% (range from 10% to 58%), while the current study found that 48 out of 263 patients (18.3%) with thyroid nodules also had concomitant Hashimoto thyroiditis. Additionally, it has been proposed that Hashimoto thyroiditis plays a dual role in the onset and progression of papillary thyroid carcinoma (PTC), raising the risk of PTC while delaying the disease’s progression (31,32). Ignoring the impact of widespread thyroid lesions in the diagnostic imaging of benign and malignant thyroid nodules might lead to subjective underdiagnosis. While previous studies have utilized vascular normalization (e.g., NIC_CCA) to reduce injection-related or hemodynamic variations (15,20), few have systematically investigated normalization against the background thyroid parenchyma (NIC_P/NZeff_P) to correct for the effects of concomitant thyroid autoimmunity (e.g., HT), baseline iodine content, or local filtration/uptake differences.

To lessen the potential impacts, we employed a dual-normalization strategy: standardizing the quantitative parameters within the lesion against both the common carotid artery (for vascular reference) and the background thyroid parenchyma. We discovered that NIC_P_IAP has a specific value in the differential diagnosis of benign and malignant nodules. According to this result, the integration of parenchyma-normalized DECT parameters can thus aid in more accurate tumor detection and diagnosis. Subsequently, a subgroup analysis of HT was conducted across the entire study population using the nomogram model. This analysis identified significant interactions between HT and several variables in the model (see Tables S4,S5). However, the investigation was constrained by the limited sample size and the constraints imposed by the available data. Future studies integrating DECT parameters with multimodal data are warranted, wherein the presented dual-normalization strategy may control heterogeneity and minimize visual and subjective bias in nodule assessment caused by variations in the number of thyroid follicular cells, basal iodine content, tumor differentiation, and microvessel densities, among other factors. Therefore, the derived variables related to iodine concentration and thyroid background need to be paid attention to in future research.

According to our most recent research, enhanced blurring and thyroid edge interruption are crucial imaging characteristics for distinguishing between benign and malignant thyroid nodules, and more likely to appear in the latter. Song et al. (20) also found that enhanced blurring signs were also applicable to the differential diagnosis of atypical benign and malignant thyroid nodule. This characteristic was seen in 96 (73.8%) of the malignant nodule tumors in our investigation. The appearance of enhanced blurring may be caused by the neovascularization of the PTC tumor margins, which is denser than the tumor center. That is the tumor margins have a relatively rich blood supply. As a result, enhancement is more pronounced than in the center, with less density difference from the peritumoral thyroid tissue (33,34). Thyroid edge interruption signs (19,35) was observed in 86 (66.2%) of the malignant nodules in our investigation, and in axial or multiplane reconstruction shows that the largest diameter of the tumor is usually located at the thyroid edge. Due to some nodules having more obvious edge enhancement after enhancement, this phenomenon partially or completely disappears, as a result the assessment of this sign requires a combination of plain scan and enhanced images.

As described in earlier correlative research, punctate calcification with a pathologic foundation in grit has a high sensitivity and specificity in the identification of malignant thyroid nodules, particularly PTC (19,20). Microcalcifications were seen by Kim et al. (36) as a hallmark of papillary thyroid cancer. Nevertheless, Yi et al. (37) in a retrospective study of 405 patients with thyroid nodules based on qualitative and quantitative parameters of DLCT, also did not include this sign in the nomogram model established to identify benign and malignant thyroid nodules. Although the possible causes cannot be accurately evaluated, the report of Ohmori
et al. (38), may supplement the deficiencies in our study. When thyroid cancer is complicated with Hashimoto thyroiditis, the number of sand bodies on the histology is reduced, indicating that PTC complications will affect the formation of microcalcification and further reduce the probability of microcalcification. Due to the limited sample size, there was no classification discussion on such cases. This makes us realize that the results of the study may be affected by the different proportion of Hashimoto thyroiditis in the samples, and there are limitations in the evaluation of benign and malignant nodules by single image morphological features or quantitative parameters. In addition, it may be affected by various factors such as nodular heterogeneity, scanning conditions, and analysis methods. Other imaging features and quantitative parameters, lymph node status, laboratory examination, etc. should be combined to reduce missed diagnosis in actual clinical practice.

As a validated first-line modality for thyroid nodule assessment, ultrasonography integrated with the Thyroid Imaging Reporting and Data System (TI-RADS) is favored for its broad accessibility, favorable cost profile, and safety without ionizing radiation (1-3). Accordingly, the DECT-radiological nomogram presented here is designed to play a complementary role to standard ultrasound and fine-needle aspiration (FNA). Its utility is most evident in clinically challenging situations, such as nodules with indeterminate ultrasound characteristics, non-diagnostic FNA cytology, or cases where preoperative comprehensive neck evaluation is indicated. By delivering synergistic anatomical and functional quantitative data, DECT can thereby offer critical supplementary evidence for multidisciplinary clinical planning. Prospective studies conducting direct head-to-head comparisons between this nomogram and TI-RADS are needed to further elucidate its integrated value within the diagnostic workflow.

Our retrospective study is a differential diagnosis model based on the commonly used CT radiological features and DECT quantitative parameters in the differentiation of benign and malignant thyroid nodules. The combined application of morphology and DECT quantitative parameters deepens the understanding of common solid thyroid benign and malignant nodules, which can effectively assist clinical decision-making. It also lays the foundation for our further research in the later stage. Our predictive nomogram can be used for more accurate diagnosis, which is of great value for pre-clinical decision-making.

Our study has the following limitations. First, the pathologic types of thyroid nodules included in our study lack diversity, and only the most common pathologic types of benign and malignant tumors are included. Therefore, our findings may not be applicable to the differential diagnosis of benign and malignant nodules with rare pathologic types. Second, this is a retrospective study from a single institution, with a small sample size and lack of external data validation, which inevitably leads to certain selection bias. Third, in our study, only arteriovenous phase images were used for quantitative analysis, and the changes in quantitative parameters of benign and malignant thyroid nodules before and after contrast agent injection were not evaluated, so the iodine uptake level of benign and malignant lesions could not be accurately evaluated. In addition, due to the limited sample size, we failed to include enough variables and clinical information, and the model may have defects in reflecting tumor heterogeneity. Therefore, it is necessary to conduct prospective multi-center studies to include more pathological subtypes and explore the clinical application value of our diagnostic model in more ‘real world’ scenarios. What cannot be overlooked is that iodine concentration and derived quantitative parameters of background thyroid tissue merit further investigation in future studies.

Conclusions

In conclusion, this study indicated that thyroid edge interruption, enhanced blurring, IC_IAP, NIC_P_IAP, and Zeff_IVP were independent predictors of malignancy in common thyroid nodules. The integrated qualitative and quantitative multi-parameter nomogram regression model has a high diagnostic performance, which can help to improve the diagnostic accuracy and can provide a more objective imaging basis for the clinic. The derived quantitative parameters obtained noninvasively by DECT can also aid in the more objective diagnosis of benign and malignant nodules.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2025-421/rc

Data Sharing Statement: Available at https://gs.amegroups.com/article/view/10.21037/gs-2025-421/dss

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2025-421/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2025-421/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The First People’s Hospital of Lianyungang (approval No. KY-20220506001-01). Informed consent was waived in this retrospective study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Durante C, Hegedüs L, Czarniecka A, et al. 2023 European Thyroid Association Clinical Practice Guidelines for thyroid nodule management. Eur Thyroid J 2023;12:e230067. [Crossref] [PubMed]
2. Durante C, Grani G, Lamartina L, et al. The Diagnosis and Management of Thyroid Nodules: A Review. JAMA 2018;319:914-24. [Crossref] [PubMed]
3. Ringel MD, Sosa JA, Baloch Z, et al. 2025 American Thyroid Association Management Guidelines for Adult Patients with Differentiated Thyroid Cancer. Thyroid 2025;35:841-985. [Crossref] [PubMed]
4. Persichetti A, Di Stasio E, Coccaro C, et al. Inter- and Intraobserver Agreement in the Assessment of Thyroid Nodule Ultrasound Features and Classification Systems: A Blinded Multicenter Study. Thyroid 2020;30:237-42. [Crossref] [PubMed]
5. Gitto S, Grassi G, De Angelis C, et al. A computer-aided diagnosis system for the assessment and characterization of low-to-high suspicion thyroid nodules on ultrasound. Radiol Med 2019;124:118-25. [Crossref] [PubMed]
6. Chen L, Zhang J, Meng L, et al. A new ultrasound nomogram for differentiating benign and malignant thyroid nodules. Clin Endocrinol (Oxf) 2019;90:351-9. [Crossref] [PubMed]
7. Chung SR, Baek JH, Choi YJ, et al. The Role of Core Needle Biopsy for the Evaluation of Thyroid Nodules with Suspicious Ultrasound Features. Korean J Radiol 2019;20:158-65. [Crossref] [PubMed]
8. Hsiao V, Massoud E, Jensen C, et al. Diagnostic Accuracy of Fine-Needle Biopsy in the Detection of Thyroid Malignancy: A Systematic Review and Meta-analysis. JAMA Surg 2022;157:1105-13. [Crossref] [PubMed]
9. Idrees T, Rashied AA, Kim B. Nondiagnostic Fine Needle Aspiration of Thyroid Nodules: Review of Predisposing Factors. Endocr Pract 2025;31:85-91. [Crossref] [PubMed]
10. Song M, Yue Y, Jin Y, et al. Intravoxel incoherent motion and ADC measurements for differentiating benign from malignant thyroid nodules: utilizing the most repeatable region of interest delineation at 3.0 T. Cancer Imaging 2020;20:9. [Crossref] [PubMed]
11. Zheng T, Zhang Y, Wang H, et al. Thyroid imaging reporting and data system with MRI morphological features for thyroid nodules: diagnostic performance and unnecessary biopsy rate. Cancer Imaging 2024;24:74. [Crossref] [PubMed]
12. Chen Y, Yu J, Lv L, et al. Predictive Value of Arterial Enhancement Fraction Derived from Dual-Layer Spectral Computed Tomography for Thyroid Microcarcinoma. Diagnostics (Basel) 2025;15:2427. [Crossref] [PubMed]
13. Hickethier T, Kroeger JR, Lennartz S, et al. Venous-phase chest CT with reduced contrast medium dose: Utilization of spectral low keV monoenergetic images improves image quality. Eur J Radiol 2020;122:108756. [Crossref] [PubMed]
14. Li F, Huang F, Liu C, et al. Parameters of dual-energy CT for the differential diagnosis of thyroid nodules and the indirect prediction of lymph node metastasis in thyroid carcinoma: a retrospective diagnostic study. Gland Surg 2022;11:913-26. [Crossref] [PubMed]
15. Ren X, Song Z, Zhang D, et al. Differentiation of benign and malignant lesions in Bethesda III and IV thyroid nodules via dual-energy computed tomography quantitative parameters and morphologic features. Quant Imaging Med Surg 2024;14:4567-78. [Crossref] [PubMed]
16. Liu X, Ouyang D, Li H, et al. Papillary thyroid cancer: dual-energy spectral CT quantitative parameters for preoperative diagnosis of metastasis to the cervical lymph nodes. Radiology 2015;275:167-76. [Crossref] [PubMed]
17. Zou Y, Sun S, Liu Q, et al. A new prediction model for lateral cervical lymph node metastasis in patients with papillary thyroid carcinoma: Based on dual-energy CT. Eur J Radiol 2021;145:110060. [Crossref] [PubMed]
18. Zhou Y, Su GY, Hu H, et al. Radiomics analysis of dual-energy CT-derived iodine maps for diagnosing metastatic cervical lymph nodes in patients with papillary thyroid cancer. Eur Radiol 2020;30:6251-62. [Crossref] [PubMed]
19. Han ZJ, Shu YY, Lai XF, et al. Value of computed tomography in determining the nature of papillary thyroid microcarcinomas: evaluation of the computed tomographic characteristics. Clin Imaging 2013;37:664-8. [Crossref] [PubMed]
20. Song Z, Li Q, Zhang D, et al. Nomogram based on spectral CT quantitative parameters and typical radiological features for distinguishing benign from malignant thyroid micro-nodules. Cancer Imaging 2023;23:13. [Crossref] [PubMed]
21. Li Q, Song Z, Zhang D, et al. Diagnostic accuracy of dual-energy computed tomography-based nomogram for differentiating papillary thyroid microcarcinomas from micronodular goiters. Quant Imaging Med Surg 2023;13:3428-40. [Crossref] [PubMed]
22. Lee DH, Lee YH, Seo HS, et al. Dual-energy CT iodine quantification for characterizing focal thyroid lesions. Head Neck 2019;41:1024-31. [Crossref] [PubMed]
23. Simons D, Kachelriess M, Schlemmer HP. Recent developments of dual-energy CT in oncology. Eur Radiol 2014;24:930-9. [Crossref] [PubMed]
24. Rassouli N, Etesami M, Dhanantwari A, et al. Detector-based spectral CT with a novel dual-layer technology: principles and applications. Insights Imaging 2017;8:589-98. [Crossref] [PubMed]
25. Barbara Krug K, Schömig-Markiefka B, Campbell GM, et al. Correlation of CT-data derived from multiparametric dual-layer CT-maps with immunohistochemical biomarkers in invasive breast carcinomas. Eur J Radiol 2022;156:110544. [Crossref] [PubMed]
26. Fan S, Li X, Zheng L, et al. Correlations between the iodine concentrations from dual energy computed tomography and molecular markers Ki-67 and HIF-1α in rectal cancer: A preliminary study. Eur J Radiol 2017;96:109-14. [Crossref] [PubMed]
27. Liu J, Pan H, Lin Q, et al. Added value of spectral parameters in diagnosing metastatic lymph nodes of pT1-2 rectal cancer. Abdom Radiol (NY) 2023;48:1260-7. [Crossref] [PubMed]
28. Chen W, Lin G, Cheng F, et al. Development and Validation of a Dual-Energy CT-Based Model for Predicting the Number of Central Lymph Node Metastases in Clinically Node-Negative Papillary Thyroid Carcinoma. Acad Radiol 2024;31:142-56. [Crossref] [PubMed]
29. Geng D, Zhou Y, Su GY, et al. Influence of sex, age and thyroid function indices on dual-energy computed tomography-derived quantitative parameters of thyroid in patients with or without Hashimoto's thyroiditis. BMC Med Imaging 2023;23:25. [Crossref] [PubMed]
30. Scheepers MHMC, Al-Difaie ZJJ, Bouvy ND, et al. Four-Dimensional Dual-Energy Computed Tomography-Derived Parameters and Their Correlation with Thyroid Gland Functional Status. Tomography 2025;11:22. [Crossref] [PubMed]
31. Janicki L, Patel A, Jendrzejewski J, et al. Prevalence and Impact of BRAF mutation in patients with concomitant papillary thyroid carcinoma and Hashimoto's thyroiditis: a systematic review with meta-analysis. Front Endocrinol (Lausanne) 2023;14:1273498. [Crossref] [PubMed]
32. Xu J, Ding K, Mu L, et al. Hashimoto's Thyroiditis: A "Double-Edged Sword" in Thyroid Carcinoma. Front Endocrinol (Lausanne) 2022;13:801925. [Crossref] [PubMed]
33. Radzina M, Ratniece M, Putrins DS, et al. Performance of Contrast-Enhanced Ultrasound in Thyroid Nodules: Review of Current State and Future Perspectives. Cancers (Basel) 2021;13:5469. [Crossref] [PubMed]
34. Li X, Gao F, Li F, et al. Qualitative analysis of contrast-enhanced ultrasound in the diagnosis of small, TR3-5 benign and malignant thyroid nodules measuring ≤1 cm. Br J Radiol 2020;93:20190923. [Crossref] [PubMed]
35. Liu Y, Liu H, Qian CL, et al. Utility of quantitative contrast-enhanced ultrasound for the prediction of extracapsular extension in papillary thyroid carcinoma. Sci Rep 2017;7:1472. [Crossref] [PubMed]
36. Kim DW, Jung SJ, Baek HJ. Computed tomography features of benign and malignant solid thyroid nodules. Acta Radiol 2015;56:1196-202. [Crossref] [PubMed]
37. Yi R, Li T, Xie G, et al. Diagnosis of benign and malignant thyroid nodules by a dual-layer spectral detector CT-based nomogram. Front Oncol 2023;13:1132817. [Crossref] [PubMed]
38. Ohmori N, Miyakawa M, Ohmori K, et al. Ultrasonographic findings of papillary thyroid carcinoma with Hashimoto's thyroiditis. Intern Med 2007;46:547-50. [Crossref] [PubMed]