Predicting Malignancy in Thyroid Nodules: Radiomics Score Versus 2017 American College of Radiology Thyroid Imaging, Reporting and Data System

A predictive nomogram of thyroid nodules based on deep learning ultrasound image analysis

Objectives

The ultrasound characteristics of benign and malignant thyroid nodules were compared to develop a deep learning model, aiming to establish a nomogram model based on deep learning ultrasound image analysis to improve the predictive performance of thyroid nodules.

Materials and methods

This retrospective study analyzed the clinical and ultrasound characteristics of 2247 thyroid nodules from March 2016 to October 2023. Among them, 1573 nodules were used for training and testing the deep learning models, and 674 nodules were used for validation, and the deep learning predicted values were obtained. These 674 nodules were randomly divided into a training set and a validation set in a 7:3 ratio to construct a nomogram model.

Results

The accuracy of the deep learning model in 674 thyroid nodules was 0.886, with a precision of 0.900, a recall rate of 0.889, and an F1-score of 0.895. The binary logistic analysis of the training set revealed that age, echogenic foci, and deep learning predicted values were statistically significant (P<0.05). These three indicators were used to construct the nomogram model, showing higher accuracy compared to the China thyroid imaging reports and data systems (C-TIRADS) classification and deep learning models. Moreover, the nomogram model exhibited high calibration and clinical benefits.

Conclusion

Age, deep learning predicted values, and echogenic foci can be used as independent predictive factors to distinguish between benign and malignant thyroid nodules. The nomogram integrates deep learning and patient clinical ultrasound characteristics, yielding higher accuracy than the application of C-TIRADS or deep learning models alone.

Role of Artificial Intelligence in Thyroid Cancer Diagnosis

The progress of artificial intelligence (AI), particularly its core algorithms-machine learning (ML) and deep learning (DL)-has been significant in the medical field, impacting both scientific research and clinical practice. These algorithms are now capable of analyzing ultrasound images, processing them, and providing outcomes, such as determining the benignity or malignancy of thyroid nodules. This integration into ultrasound machines is referred to as computer-aided diagnosis (CAD). The use of such software extends beyond ultrasound to include cytopathological and molecular assessments, enhancing the estimation of malignancy risk. AI's considerable potential in cancer diagnosis and prevention is evident. This article provides an overview of AI models based on ML and DL algorithms used in thyroid diagnostics. Recent studies demonstrate their effectiveness and diagnostic role in ultrasound, pathology, and molecular fields. Notable advancements include content-based image retrieval (CBIR), enhanced saliency CBIR (SE-CBIR), Restore-Generative Adversarial Networks (GANs), and Vision Transformers (ViTs). These new algorithms show remarkable results, indicating their potential as diagnostic and prognostic tools for thyroid pathology. The future trend points to these AI systems becoming the preferred choice for thyroid diagnostics.

Multiparameter and Ultrasound Radiomics Nomogram to Predict the Aggressiveness of Papillary Thyroid Carcinomas: A Multicenter, Retrospective Study

RATIONALE AND OBJECTIVES

To construct a multiparameter radiomics nomogram based on ultrasound (US) to predict the aggressiveness of thyroid papillary carcinoma (PTC).

MATERIALS AND METHODS

In total, 471 consecutive patients from three institutions were included in this study. Among them, patients from institution 1 were used for training (n = 294) and internal validation (n = 92), while 85 patients from institution 2 and institution 3 were used for external validation. Radiomics features were extracted from the conventional US. The least absolute shrinkage was employed to select the most relevant features for the aggressiveness of PTC, along with the maximum relevance minimum redundancy algorithm and selection operator. These features were then used to construct the radiomics signature (RS). Subsequently, relevant multiparameter ultrasound (MPUS) features from shear-wave elastic (SWE) and strain elastography (SE) will be extracted using multivariable logistic regression. The final radionics nomogram was conducted using the RS, clinical information, and conventional US and MPUS features. The receiver operating characteristic (ROC), calibration, and decision curves were used to evaluate the performance of the nomogram.

RESULTS

Multivariable logistic regression analysis indicated that age, nodule size, capsule abutment, SWV tumor, and RS were independent predictors of the aggressiveness of PTC. The radiomics nomogram, utilizing these characteristics, displayed impressive performance with an AUC of 0.920 [95% CI, 0.889-0.950], 0.901 [95% CI, 0.839-0.963], and 0.896 [95% CI, 0.823-0.969] in the training, internal, and external validation cohort. It outperformed the clinical US, MPUS, and RS models (p < 0.05). The decision curve analysis indicated that the nomogram offered valuable clinical utility.

CONCLUSION

The nomogram incorporated MPUS and radiomics have good diagnostic performance in predicting the aggressiveness of PTC which may help in the selection of the surgical modality.

Model Based on Ultrasound Radiomics and Machine Learning to Preoperative Differentiation of Follicular Thyroid Neoplasm

OBJECTIVES

To evaluate the value of radiomics based on ultrasonography in differentiating follicular thyroid carcinoma (FTC) and follicular thyroid adenoma (FTA) and construct a tool for preoperative noninvasive predicting FTC and FTA.

METHODS

The clinical data and ultrasound images of 389 patients diagnosed with FTC or FTA postoperatively were retrospectively analyzed at 3 institutions from January 2017 to December 2023. Patients in our hospital were randomly assigned in a 7:3 ratio to training cohort and validation cohort. External test cohort consisted of data collected from other 2 hospitals. Radiomics features were used to develop models based on different machine learning classifiers. A combined model was developed combining radiomics features with clinical characteristics and a nomogram was depicted. The performance of the models was assessed by area under the receiver operating characteristic curve (AUC), calibration curve and decision curve.

RESULTS

Radiomics model based on random forest showed best performance in discriminating FTC and FTA, with AUCs 0.880 (95% confidence interval [CI]: 0.8290-0.9308), 0.871 (95% CI: 0.7690-0.9734), and 0.821 (95% CI: 0.7036-0.9389) in training, validation, and test cohort, respectively. The combined model presented better efficacy comparing with clinical model and radiomics model, with AUCs 0.883 (95% CI: 0.8359-0.9295), 0.874 (95% CI: 0.7873-0.9615), and 0.876 (0.7809-0.9714) in training, validation, and test cohort, respectively. The calibration curves suggested good consistency and decision curves showed the highest overall clinical benefit for the combined model.

CONCLUSIONS

Ultrasound radiomics model based on random forest is feasible to differentiate FTC and FTA, and the combined model is an intuitively noninvasive tool for FTC and FTA preoperative identification.

Radiomics analysis of dual-energy CT-derived iodine maps for differentiating malignant from benign thyroid nodules

BACKGROUND

Many thyroid nodules are detected incidentally with the widespread use of sensitive imaging techniques; however, only a fraction of these nodules are malignant, resulting in unnecessary medical expenditures and anxiety. The major challenge is to differentiate benign thyroid nodules from malignant ones. The application of dual-energy computed tomography (DECT) and radiomics provides a new diagnostic approach. Studies applying radiomics from primary tumours on iodine maps to differentiate malignant from benign thyroid nodules are still lacking.

PURPOSE

To determine the ability of an iodine map-based radiomic nomogram in the venous phase for differentiating malignant thyroid nodules from benign nodules.

METHODS

A total of 141 patients with thyroid nodules who underwent DECT were enrolled and randomly assigned to the training and test cohorts between January 2018 and January 2019. The radiomic score (Rad-score) was derived from nine quantitative features of the iodine maps. Stepwise logistic regression analysis was used to develop radiomic, clinical and combined models. Age, normalized iodine concentration (NIC), and cyst changes were used to construct the clinical model. Receiver operating characteristic (ROC) curve analysis, sensitivity and specificity were performed to analyse the ability of the models to predict malignant thyroid nodules. Calibration analysis was used to test the fitness of the models. Decision curve analysis (DCA) and nomogram construction were also performed.

RESULTS

According to the clinical model, age (0.989 [0.984, 0.995]; p < 0.001), NIC (0.778 [0.640, 0.995]; p = 0.01), and cyst changes (0.617 [0.507, 0.751]; p < 0.001) were independently associated with malignant thyroid nodules. According to the combined model, age (0.994 [0.989, 0.999]; p = 0.01), NIC (0.797 [0.674, 0.941]; p = 0.008), cyst changes (0.786 [0.653, 0.947]; p = 0.01), and the rad-score (1.106 [1.070, 1.143]; p < 0.001) were independently associated with malignant thyroid nodules. The combined model achieved satisfactory discrimination in predicting malignant thyroid nodules and had greater predictive value in the training (AUC [areas under the curve], 0.96 vs. 0.87; p = 0.01) and test (AUC, 0.90 vs. 0.79; p = 0.04) cohorts than did the clinical model.

CONCLUSIONS

The radiomics nomogram based on iodine maps is useful to distinguish malignant thyroid nodules from benign thyroid nodules.

A Machine Learning Model Based on Thyroid US Radiomics to Discriminate Between Benign and Malignant Nodules

Background/Objectives: Thyroid nodules are a very common finding, mostly benign but sometimes malignant, and thus require accurate diagnosis. Ultrasound and fine needle biopsy are the most widely used and reliable diagnostic methods to date, but they are sometimes limited in addressing benign from malignant nodules, mainly with regard to ultrasound, by the operator's experience. Radiomics, quantitative feature extraction from medical images and machine learning offer promising avenues to improve diagnosis. The aim of this work was to develop a machine learning model based on thyroid ultrasound images to classify nodules into benign and malignant classes. Methods: For this purpose, images of ultrasonography from 142 subjects were collected. Among these subjects, 40 patients (28.2%) belonged to the class "malignant" and 102 patients (71.8%) belonged to the class "benign", according to histological diagnosis from fine-needle aspiration. This image set was used for the training, cross-validation and internal testing of three different machine learning models. A robust radiomic approach was applied, under the hypothesis that the radiomic feature could capture the disease heterogeneity among the two groups. Three models consisting of four ensembles of machine learning classifiers (random forests, support vector machines and k-nearest neighbor classifiers) were developed for the binary classification task of interest. The best performing model was then externally tested on a cohort of 21 new patients. Results: The best model (ensemble of random forest) showed Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) (%) of 85 (majority vote), 83.7 ** (mean) [80.2-87.2], accuracy (%) of 83, 81.2 ** [77.1-85.2], sensitivity (%) of 70, 67.5 ** [64.3-70.7], specificity (%) of 88, 86.5 ** [82-91], positive predictive value (PPV) (%) of 70, 66.5 ** [57.9-75.1] and negative predictive value (NPV) (%) of 88, 87.1 ** [85.5-88.8] (* p < 0.05, ** p < 0.005) in the internal test cohort. It achieved an accuracy of 90.5%, a sensitivity of 100%, a specificity of 86.7%, a PPV of 75% and an NPV of 100% in the external testing cohort. Conclusions: The model constituted of four ensembles of random forest classifiers could identify all the malignant nodes and the consistent majority of benign in the external testing cohort.

Ultrasomics differentiation of malignant and benign focal liver lesions based on contrast-enhanced ultrasound

OBJECTIVES

To establish a nomogram for differentiating malignant and benign focal liver lesions (FLLs) using ultrasomics features derived from contrast-enhanced ultrasound (CEUS).

METHODS

527 patients were retrospectively enrolled. On the training cohort, ultrasomics features were extracted from CEUS and b-mode ultrasound (BUS). Automatic feature selection and model development were performed using the Ultrasomics-Platform software, outputting the corresponding ultrasomics scores. A nomogram based on the ultrasomics scores from artery phase (AP), portal venous phase (PVP) and delayed phase (DP) of CEUS, and clinical factors were established. On the validation cohort, the diagnostic performance of the nomogram was assessed and compared with seniorexpert and resident radiologists.

RESULTS

In the training cohort, the AP, PVP and DP scores exhibited better differential performance than BUS score, with area under the curve (AUC) of 84.1-85.1% compared with the BUS (74.6%, P < 0.05). In the validation cohort, the AUC of combined nomogram and expert was significantly higher than that of the resident (91.4% vs. 89.5% vs. 79.3%, P < 0.05). The combined nomogram had a comparable sensitivity with the expert and resident (95.2% vs. 98.4% vs. 97.6%), while the expert had a higher specificity than the nomogram and the resident (80.6% vs. 72.2% vs. 61.1%, P = 0.205).

CONCLUSIONS

A CEUS ultrasomics based nomogram had an expert level performance in FLL characterization.

A Comparative Study of Two Radiomics-Based Blood Flow Modes with Thyroid Imaging Reporting and Data System in Predicting Malignancy of Thyroid Nodules and Reducing Unnecessary Fine-Needle Aspiration Rate

RATIONALE AND OBJECTIVES

We aimed to compare superb microvascular imaging (SMI)-based radiomics methods, and contrast-enhanced ultrasound (CEUS)-based radiomics methods to the American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) for classifying thyroid nodules (TNs) and reducing unnecessary fine-needle aspiration biopsy (FNAB) rate.

MATERIALS AND METHODS

This retrospective study enrolled a dataset of 472 pathologically confirmed TNs. Radiomics characteristics were extracted from B-mode ultrasound (BMUS), SMI, and CEUS images, respectively. After eliminating redundant features, four radiomics scores (Rad-scores) were constructed. Using multivariable logistic regression analysis, four radiomics prediction models incorporating Rad-score and corresponding US features were constructed and validated in terms of discrimination, calibration, decision curve analysis, and unnecessary FNAB rate.

RESULTS

The diagnostic performance of the BMUS + SMI radiomics method was better than ACR TI-RADS (area under the curve [AUC]: 0.875 vs. 0.689 for the training cohort, 0.879 vs. 0.728 for the validation cohort) (P < 0.05), and comparable with BMUS + CEUS radiomics method (AUC: 0.875 vs. 0.878 for the training cohort, 0.879 vs. 0.865 for the validation cohort) (P > 0.05). Decision curve analysis showed that the BMUS+SMI radiomics method could achieve higher net benefits than the BMUS radiomics method and ACR TI-RADS when the threshold probability was between 0.13 and 0.88 in the entire cohort. When applying the BMUS+SMI radiomics method, the unnecessary FNAB rate reduced from 43.4% to 13.9% in the training cohort and from 45.6% to 18.0% in the validation cohorts in comparison to ACR TI-RADS.

CONCLUSION

The dual-modal SMI-based radiomics method is convenient and economical and can be an alternative to the dual-modal CEUS-based radiomics method in helping radiologists select the optimal clinical strategy for TN management.

Targeted sequencing of DNA/RNA combined with radiomics predicts lymph node metastasis of papillary thyroid carcinoma

OBJECTIVE

The aim of our study is to find a better way to identify a group of papillary thyroid carcinoma (PTC) with more aggressive behaviors and to provide a prediction model for lymph node metastasis to assist in clinic practice.

METHODS

Targeted sequencing of DNA/RNA was used to detect genetic alterations. Gene expression level was measured by quantitative real-time PCR, western blotting or immunohistochemistry. CCK8, transwell assay and flow cytometry were used to investigate the effects of concomitant gene alterations in PTC. LASSO-logistics regression algorithm was used to construct a nomogram model integrating radiomic features, mutated genes and clinical characteristics.

RESULTS

172 high-risk variants and 7 fusion types were detected. The mutation frequencies in BRAF, TERT, RET, ATM and GGT1 were significantly higher in cancer tissues than benign nodules. Gene fusions were detected in 16 samples (2 at the DNA level and 14 at the RNA level). ATM mutation (ATMMUT) was frequently accompanied by BRAFMUT, TERTMUT or gene fusions. ATMMUT alone or ATM co-mutations were significantly positively correlated with lymph node metastasis. Accordingly, ATM knock-down PTC cells bearing BRAFV600E, KRASG12R or CCDC6-RET had higher proliferative ability and more aggressive potency than cells without ATM knock-down in vitro. Furthermore, combining gene alterations and clinical features significantly improved the predictive efficacy for lymph node metastasis of radiomic features, from 71.5 to 87.0%.

CONCLUSIONS

Targeted sequencing of comprehensive genetic alterations in PTC has high prognostic value. These alterations, in combination with clinical and radiomic features, may aid in predicting invasive PTC with higher accuracy.

The machine learning-based model for lateral lymph node metastasis of thyroid medullary carcinoma improved the prediction ability of occult metastasis

BACKGROUND

For medullary thyroid carcinoma (MTC) with no positive findings in the lateral neck before surgery, whether prophylactic lateral neck dissection (LND) is needed remains controversial. A better way to predict occult metastasis in the lateral neck is needed.

METHODS

From January 2010 to January 2022, patients who were diagnosed with MTC and underwent primary surgery at our hospital were retrospectively reviewed. We collected the patients' baseline characteristics, surgical procedure, and rescored the ultrasound images of the primary lesions using American College of Radiology (ACR) Thyroid Imaging, Reporting and Data System (TI-RADS). Regularized logistic regression, 5-fold cross-validation and decision curve analysis was applied for lateral lymph node metastasis (LLNM) model's development and validation. Then, we tested the predictive ability of the LLNM model for occult LLNM in cN0-1a patients.

RESULTS

A total of 218 patients were enrolled. Five baseline characteristics and two TI-RADS features were identified as high-risk factors for LLNM: gender, baseline calcitonin (Ctn), tumor size, multifocality, and central lymph node (CLN) status, as well as TI-RADS margin and level. A LLNM model was developed and showed a good discrimination with 5-fold cross-validation mean area under curve (AUC) = 0.92 ± 0.03 in the test dataset. Among cN0-1a patients, our LLNM model achieved an AUC of 0.91 (95% CI, 0.88-0.94) for predicting occult LLNM, which was significantly higher than the AUCs of baseline Ctn (0.83) and CLN status (0.64).

CONCLUSIONS

We developed a LLNM prediction model for MTC using machine learning based on clinical baseline characteristics and TI-RADS. Our model can predict occult LLNM for cN0-1a patients more accurately, then benefit the decision of prophylactic LND.

Development and Validation of a Nomogram Based on Multimodality Ultrasonography Images for Differentiating Malignant from Benign American College of Radiology Thyroid Imaging, Reporting and Data System (TI-RADS) 3-5 Thyroid Nodules

OBJECTIVE

The aim of the work described here was to develop and validate a predictive nomogram based on combined image features of gray-scale ultrasonography (US), elastosonography (ES) and contrast-enhanced US (CEUS) to differentiate malignant from benign American College of Radiology Thyroid Imaging, Reporting and Data System (ACR TI-RADS) 3-5 thyroid nodules.

METHODS

Among 2767 thyroid nodules scanned by CEUS in Xijing Hospital between April 2014 and November 2018, 669 nodules classified as ACR TI-RADS 3-5 were included, with confirmed diagnosis and ES examination. Four hundred fifty-five nodules were set as a training cohort and 214 as a validation cohort. Images were categorized as gray-scale US ACR TI-RADS 3, TI-RADS 4 and TI-RADS 5; ES patterns of ES-1 and ES-2; and CEUS patterns of either heterogeneous hypo-enhancement, concentric hypo-enhancement, homogeneous hyper-/iso-enhancement, no perfusion, hypo-enhancement with sharp margin, island-like enhancement or ring-like enhancement. On the basis of multivariate logistic regression analysis, a predictive nomogram model was developed and validated by receiver operating characteristic curve analysis.

RESULTS

In the training cohort, ACR TI-RADS 4 and 5, ES-2, heterogeneous hypo-enhancement, concentric hypo-enhancement and homogeneous hyper-/iso-enhancement were selected as predictors of malignancy by univariate logistic regression analysis. A predictive nomogram (combining indices of ACR TI-RADS, ES and CEUS) indicated excellent predictive ability for differentiating malignant from benign lesions in the training cohort: area under the receiver operating characteristic curve (AUC) = 0.93, 95% confidence interval (CI): 0.90-0.95. The prediction nomogram model was determined to have a sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of 0.84, 0.88, 0.91 and 0.81. In the validation cohort, the AUC of the prediction nomogram model was significantly higher than those of the single modalities (p < 0.005) . The AUCs of the validation cohort were 0.93 (95% CI: 0.89-0.96) and 0.93 (95% CI: 0.89-0.97), respectively, for senior and junior radiologists. The prediction nomogram model has a sensitivity, specificity, PPV and NPV of 0.86, 0.87, 0.87 and 0.86.

CONCLUSION

A predictive nomogram model combining ACR TI-RADS, ES and CEUS exhibited potential clinical utility in differentiating malignant from benign ACR TI-RADS 3-5 thyroid nodules.

Study on diagnosing thyroid nodules of ACR TI-RADS 4-5 with multimodal ultrasound radiomics technology

BACKGROUND

Explore the feasibility of using the multimodal ultrasound (US) radiomics technology to diagnose American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TI-RADS) 4-5 thyroid nodules.

METHOD

This study prospectively collected the clinical characteristics, conventional, and US elastography images of 100 patients diagnosed with ACR TI-RADS 4-5 nodules from May 2022 to 2023. Independent risk factors for malignant thyroid nodules were extracted and screened using methods such as the least absolute shrinkage and selection operator (LASSO) logistic regression (LR) model, and a multimodal US radiomics combined diagnostic model was established. Using a multifactorial LR analysis and a Rad-score rating, the predictive performance was validated and evaluated, and the final threshold range was determined to assess the clinical net benefit of the model.

RESULTS

In the training set, the US radiomics combined predictive model area under curve (AUC = 0.928) had higher diagnostic performance compared with clinical characteristics (AUC = 0.779), conventional US (AUC = 0.794), and US elastography model (AUC = 0.852). In the validation set, the multimodal US radiomics combined diagnostic model (AUC = 0.829) also had higher diagnostic performance compared with clinical characteristics (AUC = 0.799), conventional US (AUC = 0.802), and US elastography model (AUC = 0.718).

CONCLUSION

Multi-modal US radiomics technology can effectively diagnose thyroid nodules of ACR TI-RADS 4-5, and the combination of radiomics signature and conventional US features can further improve the diagnostic performance.

Future Directions in the Treatment of Thyroid and Parathyroid Disease

The surgical management of thyroid and parathyroid disease has evolved considerably since the era of Theodor Kocher. We review the current trends in thyroid and parathyroid surgery concerning robotic surgery for remote access, the use of parathyroid autofluorescence detection technology to aid in the prevention of hypocalcemia as well as the use of thermal ablation to target thyroid nodules in a minimally invasive way. We also discuss how artificial intelligence is being used to improve the workflow and diagnostics preoperatively as well as for intraoperative decision-making. We also discuss potential areas where future research may enhance outcomes.

Dual-modal radiomics nomogram based on contrast-enhanced ultrasound to improve differential diagnostic accuracy and reduce unnecessary biopsy rate in ACR TI-RADS 4-5 thyroid nodules

BACKGROUND

American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TI-RADS, TR) 4 and 5 thyroid nodules (TNs) demonstrate much more complicated and overlapping risk characteristics than TR1-3 and have a rather wide range of malignancy possibilities (> 5%), which may cause overdiagnosis or misdiagnosis. This study was designed to establish and validate a dual-modal ultrasound (US) radiomics nomogram integrating B-mode ultrasound (BMUS) and contrast-enhanced ultrasound (CEUS) imaging to improve differential diagnostic accuracy and reduce unnecessary fine needle aspiration biopsy (FNAB) rates in TR 4-5 TNs.

METHODS

A retrospective dataset of 312 pathologically confirmed TR4-5 TNs from 269 patients was collected for our study. Data were randomly divided into a training dataset of 219 TNs and a validation dataset of 93 TNs. Radiomics characteristics were derived from the BMUS and CEUS images. After feature reduction, the BMUS and CEUS radiomics scores (Rad-score) were built. A multivariate logistic regression analysis was conducted incorporating both Rad-scores and clinical/US data, and a radiomics nomogram was subsequently developed. The performance of the radiomics nomogram was evaluated using calibration, discrimination, and clinical usefulness, and the unnecessary FNAB rate was also calculated.

RESULTS

BMUS Rad-score, CEUS Rad-score, age, shape, margin, and enhancement direction were significant independent predictors associated with malignant TR4-5 TNs. The radiomics nomogram involving the six variables exhibited excellent calibration and discrimination in the training and validation cohorts, with an AUC of 0.873 (95% CI, 0.821-0.925) and 0.851 (95% CI, 0.764-0.938), respectively. The marked improvements in the net reclassification index and integrated discriminatory improvement suggested that the BMUS and CEUS Rad-scores could be valuable indicators for distinguishing benign from malignant TR4-5 TNs. Decision curve analysis demonstrated that our developed radiomics nomogram was an instrumental tool for clinical decision-making. Using the radiomics nomogram, the unnecessary FNAB rate decreased from 35.3 to 14.5% in the training cohort and from 41.5 to 17.7% in the validation cohorts compared with ACR TI-RADS.

CONCLUSION

The dual-modal US radiomics nomogram revealed superior discrimination accuracy and considerably decreased unnecessary FNAB rates in benign and malignant TR4-5 TNs. It could guide further examination or treatment options.

Modification of size cutoff for biopsy based on the American College of Radiology Thyroid Imaging Reporting and Data System (TI-RADS) for thyroid nodules in patients younger than 19 years

OBJECTIVES

To modify the size cutoff for biopsy for thyroid nodules in patients < 19 years based on the American College of Radiology Thyroid Imaging Reporting and Data System (TI-RADS) and evaluate the performance of the new criteria in two referral centers.

METHODS

Patients < 19 years with cytopathologic or surgical pathology results were retrospectively identified from two centers from May 2005 to August 2022. Patients from one center were classified as the training cohort, and those from the other center were classified as the validation cohort. The diagnostic performance, unnecessary biopsy rates, and missed malignancy rates of the TI-RADS guideline, and the new criteria (≥ 35 mm for TR3 and no threshold for TR5) were compared.

RESULTS

A total of 236 nodules from 204 patients in the training cohort and 225 nodules from 190 patients in the validation cohort were analyzed. The area under the receiver operating characteristic curve of the new criteria in identifying thyroid malignant nodules was higher (0.809 vs. 0.681, p < 0.001; 0.819 vs. 0.683, p < 0.001), and the unnecessary biopsy rates (45.0% vs. 56.8%; 42.2% vs. 56.8%) and missed malignancy rates (5.7% vs. 18.6%; 9.2% vs. 21.5%) were lower than that of the TI-RADS guideline in the training cohort and validation cohort, respectively.

CONCLUSIONS

The new criteria (≥ 35 mm for TR3 and no threshold for TR5) for biopsy based on the TI-RADS may help improve the diagnostic performance and reduce unnecessary biopsy rates and missed malignancy rates for thyroid nodules in patients < 19 years.

CLINICAL RELEVANCE STATEMENT

The study developed and validated the new criteria (≥ 35 mm for TR3 and no threshold for TR5) to indicate FNA based on the ACR TI-RADS of thyroid nodules in patients younger than 19 years.

KEY POINTS

•The AUC of the new criteria (≥ 35 mm for TR3 and no threshold for TR5) in identifying thyroid malignant nodules was higher than that of the TI-RADS guideline (0.809 vs. 0.681) in patients < 19 years. •The unnecessary biopsy rates and missed malignancy rates of the new criteria (≥ 35 mm for TR3 and no threshold for TR5) in identifying thyroid malignant nodules were lower than that of the TI-RADS guideline in patients < 19 years (45.0% vs. 56.8% and 5.7% vs. 18.6%, respectively).

Radiomics-based ultrasound models for thyroid nodule differentiation in Hashimoto's thyroiditis

Background

Previous models for differentiating benign and malignant thyroid nodules(TN) have predominantly focused on the characteristics of the nodules themselves, without considering the specific features of the thyroid gland(TG) in patients with Hashimoto's thyroiditis(HT). In this study, we analyzed the clinical and ultrasound radiomics(USR) features of TN in patients with HT and constructed a model for differentiating benign and malignant nodules specifically in this population.

Methods

We retrospectively collected clinical and ultrasound data from 227 patients with TN and concomitant HT(161 for training, 66 for testing). Two experienced sonographers delineated the TG and TN regions, and USR features were extracted using Python. Lasso regression and logistic analysis were employed to select relevant USR features and clinical data to construct the model for differentiating benign and malignant TN. The performance of the model was evaluated using area under the curve(AUC), calibration curves, and decision curve analysis(DCA).

Results

A total of 1,162 USR features were extracted from TN and the TG in the 227 patients with HT. Lasso regression identified 14 features, which were used to construct the TN score, TG score, and TN+TG score. Univariate analysis identified six clinical predictors: TI-RADS, echoic type, aspect ratio, boundary, calcification, and thyroid function. Multivariable analysis revealed that incorporating USR scores improved the performance of the model for differentiating benign and malignant TN in patients with HT. Specifically, the TN+TG score resulted in the highest increase in AUC(from 0.83 to 0.94) in the clinical prediction model. Calibration curves and DCA demonstrated higher accuracy and net benefit for the TN+TG+clinical model.

Conclusion

USR features of both the TG and TN can be utilized for differentiating benign and malignant TN in patients with HT. These findings highlight the importance of considering the entire TG in the evaluation of TN in HT patients, providing valuable insights for clinical decision-making in this population.

Predicting Malignancy of Thyroid Micronodules: Radiomics Analysis Based on Two Types of Ultrasound Elastography Images

RATIONALE AND OBJECTIVES

To develop a multimodal ultrasound radiomics nomogram for accurate classification of thyroid micronodules.

MATERIALS AND METHODS

A retrospective study including 181 thyroid micronodules within 179 patients was conducted. Radiomics features were extracted from strain elastography (SE), shear wave elastography (SWE) and B-mode ultrasound (BMUS) images. Minimum redundancy maximum relevance and least absolute shrinkage and selection operator algorithms were used to select malignancy-related features. BMUS, SE, and SWE radiomics scores (Rad-scores) were then constructed. Multivariable logistic regression was conducted using radiomics signatures along with clinical data, and a nomogram was ultimately established. The calibration, discriminative, and clinical usefulness were considered to evaluate its performance. A clinical prediction model was also built using independent clinical risk factors for comparison.

RESULTS

An aspect ratio ≥ 1, mean elasticity index, BMUS Rad-score, SE Rad-score, and SWE Rad-score were identified as the independent predictors for predicting malignancy of thyroid micronodules by multivariable logistic regression. The radiomics nomogram based on these characteristics showed favorable calibration and discriminative capabilities (AUCs: 0.903 and 0.881 for training and validation cohorts, respectively), all outperforming clinical prediction model (AUCs: 0.791 and 0.626, respectively). The decision curve analysis also confirmed clinical usefulness of the nomogram. The significant improvement of net reclassification index and integrated discriminatory improvement indicated that multimodal ultrasound radiomics signatures might work as new imaging markers for classifying thyroid micronodules.

CONCLUSION

The nomogram combining multimodal ultrasound radiomics features and clinical factors has the potential to be used for accurate diagnosis of thyroid micronodules in the clinic.

A multiparametric clinic-ultrasomics nomogram for predicting extremity soft-tissue tumor malignancy: a combined retrospective and prospective bicentric study

OBJECTIVE

We aimed at building and testing a multiparametric clinic-ultrasomics nomogram for prediction of malignant extremity soft-tissue tumors (ESTTs).

MATERIALS AND METHODS

This combined retrospective and prospective bicentric study assessed the performance of the multiparametric clinic-ultrasomics nomogram to predict the malignancy of ESTTs, when compared with a conventional clinic-radiologic nomogram. A dataset of grayscale ultrasound (US), color Doppler flow imaging (CDFI), and elastography images for 209 ESTTs were retrospectively enrolled from one hospital, and divided into the training and validation cohorts. A multiparametric ultrasomics signature was built based on multimodal ultrasomic features extracted from the grayscale US, CDFI, and elastography images of ESTTs in the training cohort. Another conventional radiologic score was built based on multimodal US features as interpreted by two experienced radiologists. Two nomograms that integrated clinical risk factors and the multiparameter ultrasomics signature or conventional radiologic score were respectively developed. Performance of the two nomograms was validated in the retrospective validation cohort, and tested in a prospective dataset of 51 ESTTs from the second hospital.

RESULTS

The multiparametric ultrasomics signature was built based on seven grayscale ultrasomic features, three CDFI ultrasomic features, and one elastography ultrasomic feature. The conventional radiologic score was built based on five multimodal US characteristics. Predictive performance of the multiparametric clinic-ultrasomics nomogram was superior to that of the conventional clinic-radiologic nomogram in the training (area under the receiver operating characteristic curve [AUC] 0.970 vs. 0.890, p = 0.006), validation (AUC: 0.946 vs. 0.828, p = 0.047) and test (AUC: 0.934 vs. 0.842, p = 0.040) cohorts, respectively. Decision curve analysis of combined training, validation and test cohorts revealed that the multiparametric clinic-ultrasomics nomogram had a higher overall net benefit than the conventional clinic-radiologic model.

CONCLUSION

The multiparametric clinic-ultrasomics nomogram can accurately predict the malignancy of ESTTs.

Ultrasonography-based radiomics and computer-aided diagnosis in thyroid nodule management: performance comparison and clinical strategy optimization

Objectives

To compare ultrasonography (US) feature-based radiomics and computer-aided diagnosis (CAD) models for predicting malignancy in thyroid nodules, and to evaluate their utility for thyroid nodule management.

Methods

This prospective study included 262 thyroid nodules obtained between January 2022 and June 2022. All nodules previously underwent standardized US image acquisition, and the nature of the nodules was confirmed by the pathological results. The CAD model exploited two vertical US images of the thyroid nodule to differentiate the lesions. The least absolute shrinkage and operator algorithm (LASSO) was applied to choose radiomics features with excellent predictive properties for building a radiomics model. Ultimately, the area under the receiver operating characteristic curve (AUC) and calibration curves were assessed to compare diagnostic performance between the models. DeLong's test was used to analyze the difference between groups. Both models were used to revise the American College of Radiology Thyroid Imaging Reporting and Data Systems (ACR TI-RADS) to provide biopsy recommendations, and their performance was compared with the original recommendations.

Results

Of the 262 thyroid nodules, 157 were malignant, and the remaining 105 were benign. The diagnostic performance of radiomics, CAD, and ACR TI-RADS models had an AUC of 0.915 (95% confidence interval (CI): 0.881-0.947), 0.814 (95% CI: 0.766-0.863), and 0.849 (95% CI: 0.804-0.894), respectively. DeLong's test showed a statistically significant between the AUC values of models (p < 0.05). Calibration curves showed good agreement in each model. When both models were applied to revise the ACR TI-RADS, our recommendations significantly improved the performance. The revised recommendations based on radiomics and CAD showed an increased sensitivity, accuracy, positive predictive value, and negative predictive value, and decreased unnecessary fine-needle aspiration rates. Furthermore, the radiomics model's improvement scale was more pronounced (33.3-16.7% vs. 33.3-9.7%).

Conclusion

The radiomics strategy and CAD system showed good diagnostic performance for discriminating thyroid nodules and could be used to optimize the ACR TI-RADS recommendation, which successfully reduces unnecessary biopsies, especially in the radiomics model.

Predicting malignancy in thyroid nodules based on conventional ultrasound and elastography: the value of predictive models in a multi-center study

BACKGROUND

This study aimed to establish predictive models based on features of Conventional Ultrasound (CUS) and elastography in a multi-center study to determine appropriate preoperative diagnosis of malignancy in thyroid nodules with different risk stratification based on 2017 Thyroid Imaging Reporting and Data System by the American College of Radiology (ACR TI-RADS) guidelines.

METHODS

Five hundred forty-eight thyroid nodules from three centers pathologically confirmed by the cytology or histology were retrospectively enrolled in the study, which were examined by CUS and elastography before fine needle aspiration (FNA) and surgery. Characteristics of CUS of thyroid nodules were reviewed according to 2017 ACR TI-RADS. Binary logistic regression analysis was used to develop the prediction models based on the different risk stratification of CUS features and elastography which were statistically significant. Values of predictive models were evaluated regarding the discrimination and calibration.

RESULTS

Binary logistic regression showed that patients' age, taller-than-wider, lobulated or irregular boundary, extra-thyroid extension, microcalcification and the elastic parameter of Virtual touch tissue imaging quantification (VTIQ) max were independent predictors for thyroid malignancy (p < 0.05) in the ACR model and showed the area under the curve (AUC) in training (0.912) and validation cohort (internal and external: 0.877 vs 0.935). Predictive models showed predictors in ACR TR4 and TR5 for malignancy and diagnostic performance of AUC in training, internal and external validation cohort respectively: the VTIQ max (p < 0.001) with AUC of 0.809 vs 0.842 vs 0.705 and the age, taller than wide, VTIQ max variables with AUC of 0.859 vs 0.830 vs 0.906 in validation cohort. All predictive models have better calibration capabilities (p > 0.05).

CONCLUSIONS

Predictive models combined CUS and elastography features would aid clinicians to make appropriate preoperative diagnosis of thyroid nodules among different risk stratification. The elastography parameter of VTIQ max has the priority in distinguishing thyroid malignancy with moderately suspicious (ACR TR4).

Radiologists with assistance of deep learning can achieve overall accuracy of benign-malignant differentiation of musculoskeletal tumors comparable with that of pre-surgical biopsies in the literature

PURPOSE

The purpose of this study was to assess if radiologists assisted by deep learning (DL) algorithms can achieve diagnostic accuracy comparable to that of pre-surgical biopsies in benign-malignant differentiation of musculoskeletal tumors (MST).

METHODS

We first conducted a systematic review of literature to get the respective overall diagnostic accuracies of fine-needle aspiration biopsy (FNAB) and core needle biopsy (CNB) in differentiating between benign and malignant MST, by synthesizing data from the articles meeting our inclusion criteria. To compared against the accuracies reported in literature, we then invited 4 radiologists, respectively with 2 (A), 6 (B), 7 (C), and 33 (D) years of experience in interpreting musculoskeletal MRI to perform diagnostic tests on our own dataset (n = 62), with and without assistance of a previously developed DL algorithm. The gold standard for benign-malignant differentiation was histopathologic confirmation or clinical/radiographic follow-up.

RESULTS

For FNAB, a meta-analysis containing 4604 samples met the inclusion criteria, with the overall diagnostic accuracy reported to be 0.77. For CNB, an overall accuracy of 0.86 was derived by synthesizing results from 7 original research articles containing a total of 587 samples. On our internal MST dataset, the invited radiologists, respectively, achieved diagnostic accuracies of 0.84 (A), 0.89 (B), 0.87 (C), and 0.90 (D), with the assistance of DL.

CONCLUSION

Use of DL algorithms on musculoskeletal dynamic contrast-enhanced MRI improved the benign-malignant differentiation accuracy of radiologists to a level comparable to that of pre-surgical biopsies. The developed DL algorithms have a potential to lower the risk of miss-diagnosing malignancy in radiological practice.

Radiomics Analysis of Gray-Scale Ultrasonographic Images of Papillary Thyroid Carcinoma > 1 cm: Potential Biomarker for the Prediction of Lymph Node Metastasis

Purpose

This study aimed to investigate radiomics analysis of ultrasonographic images to develop a potential biomarker for predicting lymph node metastasis in papillary thyroid carcinoma (PTC) patients.

Materials and Methods

This study included 431 PTC patients from August 2013 to May 2014 and classified them into the training and validation sets. A total of 730 radiomics features, including texture matrices of gray-level co-occurrence matrix and gray-level run-length matrix and single-level discrete two-dimensional wavelet transform and other functions, were obtained. The least absolute shrinkage and selection operator method was used for selecting the most predictive features in the training data set.

Results

Lymph node metastasis was associated with the radiomics score (p < 0.001). It was also associated with other clinical variables such as young age (p = 0.007) and large tumor size (p = 0.007). The area under the receiver operating characteristic curve was 0.687 (95% confidence interval: 0.616-0.759) for the training set and 0.650 (95% confidence interval: 0.575-0.726) for the validation set.

Conclusion

This study showed the potential of ultrasonography-based radiomics to predict cervical lymph node metastasis in patients with PTC; thus, ultrasonography-based radiomics can act as a biomarker for PTC.

The progress of radiomics in thyroid nodules

Due to the development of Artificial Intelligence (AI), Machine Learning (ML), and the improvement of medical imaging equipment, radiomics has become a popular research in recent years. Radiomics can obtain various quantitative features from medical images, highlighting the invisible image traits and significantly enhancing the ability of medical imaging identification and prediction. The literature indicates that radiomics has a high potential in identifying and predicting thyroid nodules. So in this article, we explain the development, definition, and workflow of radiomics. And then, we summarize the applications of various imaging techniques in identifying benign and malignant thyroid nodules, predicting invasiveness and metastasis of thyroid lymph nodes, forecasting the prognosis of thyroid malignancies, and some new advances in molecular level and deep learning. The shortcomings of this technique are also summarized, and future development prospects are provided.

Ultrasound-based radiomics score for pre-biopsy prediction of prostate cancer to reduce unnecessary biopsies

BACKGROUND

Patients undergoing prostate biopsies (PBs) suffer from low positive rates and potential risk for complications. This study aimed to develop and validate an ultrasound (US)-based radiomics score for pre-biopsy prediction of prostate cancer (PCa) and subsequently reduce unnecessary PBs.

METHODS

Between December 2015 and March 2018, 196 patients undergoing initial transrectal ultrasound (TRUS)-guided PBs were retrospectively enrolled and randomly assigned to the training or validation cohort at a ratio of 7:3. A total of 1044 radiomics features were extracted from grayscale US images of each prostate nodule. After feature selection through the least absolute shrinkage and selection operator (LASSO) regression model, the radiomics score was developed from the training cohort. The prediction nomograms were developed using multivariate logistic regression analysis based on the radiomics score and clinical risk factors. The performance of the nomograms was assessed and compared in terms of discrimination, calibration, and clinical usefulness.

RESULTS

The radiomics score consisted of five selected features. Multivariate logistic regression analysis demonstrated that the radiomics score, age, total prostate-specific antigen (tPSA), and prostate volume were independent factors for prediction of PCa (all p < 0.05). The integrated nomogram incorporating the radiomics score and three clinical risk factors reached an area under the curve (AUC) of 0.835 (95% confidence interval [CI], 0.729-0.941), thereby outperforming the clinical nomogram which based on only clinical factors and yielded an AUC of 0.752 (95% CI, 0.618-0.886) (p = 0.04). Both nomograms showed good calibration. Decision curve analysis indicated that using the integrated nomogram would add more benefit than using the clinical nomogram.

CONCLUSION

The radiomics score was an independent factor for pre-biopsy prediction of PCa. Addition of the radiomics score to the clinical nomogram shows incremental prognostic value and may help clinicians make precise decisions to reduce unnecessary PBs.

Self-supervised multi-modal fusion network for multi-modal thyroid ultrasound image diagnosis

Ultrasound is a typical non-invasive diagnostic method often used to detect thyroid cancer lesions. However, due to the limitations of the information provided by ultrasound images, shear wave elastography (SWE) and color doppler ultrasound (CDUS) are also used clinically to assist in diagnosis, which makes the diagnosis time-consuming, labor-intensive, and highly subjective process. Therefore, automatic diagnosis of benign and malignant thyroid nodules is beneficial for the clinical diagnosis of the thyroid. To this end, based on three modalities of gray-scale ultrasound images(US), SWE, and CDUS, we propose a deep learning-based multi-modal feature fusion network for the automatic diagnosis of thyroid disease based on the ultrasound images. First, three ResNet18s initialized by self-supervised learning are used as branches to extract the image information of each modality, respectively. Then, a multi-modal multi-head attention branch is used to remove the common information of three modalities, and the knowledge of each modal is combined for thyroid diagnosis. At the same time, to better integrate the features between modalities, a multi-modal feature guidance module is also proposed to guide the feature extraction of each branch and reduce the difference between each-modal feature. We verify the multi-modal thyroid ultrasound image diagnosis method on the self-collected dataset, and the results prove that this method could provide fast and accurate assistance for sonographers in diagnosing thyroid nodules.

Ultrasound Radiomics Nomogram to Diagnose Sub-Centimeter Thyroid Nodules Based on ACR TI-RADS

The aim of the present study was to develop a radiomics nomogram to assess whether thyroid nodules (TNs) < 1 cm are benign or malignant. From March 2021 to March 2022, 156 patients were admitted to the Affiliated Hospital of Nantong University, and from September 2017 to March 2022, 116 patients were retrospectively collected from the Jiangsu Provincial Hospital of Integrated Traditional Chinese and Western Medicine. These patients were divided into a training group and an external test group. A radiomics nomogram was established using multivariate logistics regression analysis using the radiomics score and clinical data, including the ultrasound feature scoring terms from the thyroid imaging reporting and data system (TI-RADS). The radiomics nomogram incorporated the correlated predictors, and compared with the clinical model (training set AUC: 0.795; test set AUC: 0.783) and radiomics model (training set AUC: 0.774; test set AUC: 0.740), had better discrimination performance and correction effects in both the training set (AUC: 0.866) and the test set (AUC: 0.866). Both the decision curve analysis and clinical impact curve showed that the nomogram had a high clinical application value. The nomogram constructed based on TI-RADS and radiomics features had good results in predicting and distinguishing benign and malignant TNs < 1 cm.

Artificial Intelligence for Thyroid Nodule Characterization: Where Are We Standing?

Simple Summary

In the present review, an up-to-date summary of the state of the art of artificial intelligence (AI) implementation for thyroid nodule characterization and cancer is provided. The opinion on the real effectiveness of AI systems remains controversial. Taking into consideration the largest and most scientifically valid studies, it is possible to state that AI provides results that are comparable or inferior to expert ultrasound specialists and radiologists. Promising data approve AI as a support tool and simultaneously highlight the need for a radiologist supervisory framework for AI provided results. Therefore, current solutions might be more suitable for educational purposes.

Abstract

Machine learning (ML) is an interdisciplinary sector in the subset of artificial intelligence (AI) that creates systems to set up logical connections using algorithms, and thus offers predictions for complex data analysis. In the present review, an up-to-date summary of the current state of the art regarding ML and AI implementation for thyroid nodule ultrasound characterization and cancer is provided, highlighting controversies over AI application as well as possible benefits of ML, such as, for example, training purposes. There is evidence that AI increases diagnostic accuracy and significantly limits inter-observer variability by using standardized mathematical algorithms. It could also be of aid in practice settings with limited sub-specialty expertise, offering a second opinion by means of radiomics and computer-assisted diagnosis. The introduction of AI represents a revolutionary event in thyroid nodule evaluation, but key issues for further implementation include integration with radiologist expertise, impact on workflow and efficiency, and performance monitoring.

Artificial Neural Network-Based Ultrasound Radiomics Can Predict Large-Volume Lymph Node Metastasis in Clinical N0 Papillary Thyroid Carcinoma Patients

Objective

To evaluate the ability of artificial neural network- (ANN-) based ultrasound radiomics to predict large-volume lymph node metastasis (LNM) preoperatively in clinical N0 disease (cN0) papillary thyroid carcinoma (PTC) patients.

Methods

From January 2020 to April 2021, 306 cN0 PTC patients admitted to our hospital were retrospectively reviewed and divided into a training (n = 183) cohort and a validation cohort (n = 123) in a 6 : 4 ratio. Radiomic features quantitatively extracted from ultrasound images were pruned to train one ANN-based radiomic model and three conventional machine learning-based classifiers in the training cohort. Furthermore, an integrated model using ANN was constructed for better prediction. Meanwhile, the prediction of the two models was evaluated in the papillary thyroid microcarcinoma (PTMC) and conventional papillary thyroid cancer (CPTC) subgroups.

Results

The radiomic model showed better discrimination than other classifiers for large-volume LNM in the validation cohort, with an area under the receiver operating characteristic curve (AUROC) of 0.856 and an area under the precision-recall curve (AUPR) of 0.381. The performance of the integrated model was better, with an AUROC of 0.910 and an AUPR of 0.463. According to the calibration curve and decision curve analysis, the radiomic and integrated models had good calibration and clinical usefulness. Moreover, the models had good predictive performance in the PTMC and CPTC subgroups.

Conclusion

ANN-based ultrasound radiomics could be a potential tool to predict large-volume LNM preoperatively in cN0 PTC patients.

Radiomics Features of Different Sizes of Medullary Thyroid Carcinoma (MTC) and Papillary Thyroid Carcinoma (PTC) Tumors: A Comparative Study

Background:

Radiomics strategies exhibit great promise in the context of thyroid nodule diagnosis. This study aimed to compare radiomics features of different sizes of medullary thyroid carcinoma (MTC) and papillary thyroid carcinoma (PTC) tumors and to compare the efficiency of radiomics approaches as a means of differentiating between these tumor types.

Methods:

In total, 86 MTC and 330 PTC nodules were divided into the macronodular (>10 mm) and micronodular (⩽10 mm) categories. The radiomics features of these nodules were analyzed to identify independent prognosis factors and evaluate the efficacy of individual and combined indicators as predictors of tumor type.

Results:

In total, 12 radiomics features were found to differ significantly between MTC and PTC macronodules, while 6 differed significantly between MTC and PTC micronodules. Shape 2D_Sphericity, firstorder_Skewness, glrlm_RunLengthNonUniformity, glszm_GrayLevelNonUniformity, and glszm_SizeZoneNonUniformity were features that were independently associated with the differential diagnoses of MTC and PTC macronodules. Receiver operating characteristic (ROC) curve analyses of the efficacy of these 5 single indicators and a combined indicator composed thereof yielded area under the curve (AUC) values of 0.621, 0.678, 0.704, 0.762, 0.747, and 0.824, respectively, with respective sensitivities of 55.3%, 43.0%, 53.1%, 56.3%, 46.9%, and 65.6%, and respective specificity values of 65.6%, 89.1%, 81.6%, 88.8%, 95.0%, and 91.1%. The glrlm_RunEntropy and glszm_SizeZoneNonUniformity features were identified as independent factors associated with the differential diagnoses of MTC and PTC micronodules. Receiver operating characteristic curve analyses of the efficacy of these 2 single indicators and a combined indicator composed thereof yielded respective AUC values of 0.678, 0.678, and 0.771; Sensitivities of 57.0%, 72.7%, and 72.7%; and specificities of 77.3%, 64.2%, and 77.5%.

Conclusions:

A range of different radiomics features can enable effective differentiation between MTC and PTC nodules of different sizes. Moreover, analyses of combinations of radiomics features yielded diagnostic efficiency values higher than those associated with single radiomics features, highlighting a more reliable approach to diagnosing MTC and PTC tumors.

A machine-learning algorithm for distinguishing malignant from benign indeterminate thyroid nodules using ultrasound radiomic features

Background: Ultrasound (US)-guided fine needle aspiration (FNA) cytology is the gold standard for the evaluation of thyroid nodules. However, up to 30% of FNA results are indeterminate, requiring further testing. In this study, we present a machine-learning analysis of indeterminate thyroid nodules on ultrasound with the aim to improve cancer diagnosis. Methods: Ultrasound images were collected from two institutions and labeled according to their FNA (F) and surgical pathology (S) diagnoses [malignant (M), benign (B), and indeterminate (I)]. Subgroup breakdown (FS) included: 90 BB, 83 IB, 70 MM, and 59 IM thyroid nodules. Margins of thyroid nodules were manually annotated, and computerized radiomic texture analysis was conducted within tumor contours. Initial investigation was conducted using five-fold cross-validation paradigm with a two-class Bayesian artificial neural networks classifier, including stepwise feature selection. Testing was conducted on an independent set and compared with a commercial molecular testing platform. Performance was evaluated using receiver operating characteristic analysis in the task of distinguishing between malignant and benign nodules. Results: About 1052 ultrasound images from 302 thyroid nodules were used for radiomic feature extraction and analysis. On the training/validation set comprising 263 nodules, five-fold cross-validation yielded area under curves (AUCs) of 0.75 [Standard Error (SE) = 0.04; P < 0.001 ] and 0.67 (SE = 0.05; P = 0.0012 ) for the classification tasks of MM versus BB, and IM versus IB, respectively. On an independent test set of 19 IM/IB cases, the algorithm for distinguishing indeterminate nodules yielded an AUC value of 0.88 (SE = 0.09; P < 0.001 ), which was higher than the AUC of a commercially available molecular testing platform (AUC = 0.81, SE = 0.11; P < 0.005 ). Conclusion: Machine learning of computer-extracted texture features on gray-scale ultrasound images showed promising results classifying indeterminate thyroid nodules according to their surgical pathology.

Risk stratification of indeterminate thyroid nodules using ultrasound and machine learning algorithms

BACKGROUND

Indeterminate thyroid nodules (Bethesda III) are challenging to characterize without diagnostic surgery. Auxiliary strategies including molecular analysis, machine learning models, and ultrasound grading with Thyroid Imaging, Reporting and Data System (TI-RADS) can help to triage accordingly, but further refinement is needed to prevent unnecessary surgeries and increase positive predictive values.

DESIGN

Retrospective review of 88 patients with Bethesda III nodules who had diagnostic surgery with final pathological diagnosis.

MEASUREMENTS

Each nodule was retrospectively scored through TI-RADS. Two deep learning models were tested, one previously developed and trained on another data set, mainly containing determinate cases and then validated on our data set while the other one trained and tested on our data set (indeterminate cases).

RESULTS

The mean TI-RADS score was 3 for benign and 4 for malignant nodules (p = .0022). Radiological high risk (TI-RADS 4,5) and low risk (TI-RADS 2,3) categories were established. The PPV for the high radiological risk category in those with >10 mm nodules was 85% (CI: 70%-93%). The NPV for low radiological risk in patients >60 years (mean age was 100% (CI: 83%-100%). The area under the curve (AUC) value of our novel classifier was 0.75 (CI: 0.62-0.84) and differed significantly from the chance-level (p < .00001).

CONCLUSIONS

Novel radiomic and radiologic strategies can be employed to assist with preoperative diagnosis of indeterminate thyroid nodules.

Radiomic Detection of Malignancy within Thyroid Nodules Using Ultrasonography-A Systematic Review and Meta-Analysis

Background: Despite investigation, 95% of thyroid nodules are ultimately benign. Radiomics is a field that uses radiological features to inform individualized patient care. We aimed to evaluate the diagnostic utility of radiomics in classifying undetermined thyroid nodules into benign and malignant using ultrasonography (US). Methods: A diagnostic test accuracy systematic review and meta-analysis was performed in accordance with PRISMA guidelines. Sensitivity, specificity, and area under curve (AUC) delineating benign and malignant lesions were recorded. Results: Seventy-five studies including 26,373 patients and 46,175 thyroid nodules met inclusion criteria. Males accounted for 24.6% of patients, while 75.4% of patients were female. Radiomics provided a pooled sensitivity of 0.87 (95% CI: 0.86−0.87) and a pooled specificity of 0.84 (95% CI: 0.84−0.85) for characterizing benign and malignant lesions. Using convolutional neural network (CNN) methods, pooled sensitivity was 0.85 (95% CI: 0.84−0.86) and pooled specificity was 0.82 (95% CI: 0.82−0.83); significantly lower than studies using non-CNN: sensitivity 0.90 (95% CI: 0.89−0.90) and specificity 0.88 (95% CI: 0.87−0.89) (p < 0.05). The diagnostic ability of radiologists and radiomics were comparable for both sensitivity (OR 0.98) and specificity (OR 0.95). Conclusions: Radiomic analysis using US provides a reproducible, reliable evaluation of undetermined thyroid nodules when compared to current best practice.

Multimodality MRI-based radiomics for aggressiveness prediction in papillary thyroid cancer

OBJECTIVE

To investigate the ability of a multimodality MRI-based radiomics model in predicting the aggressiveness of papillary thyroid carcinoma (PTC).

METHODS

This study included consecutive patients who underwent neck magnetic resonance (MR) scans and subsequent thyroidectomy during the study period. The pathological diagnosis of thyroidectomy specimens was the gold standard to determine the aggressiveness. Thyroid nodules were manually segmented on three modal MR images, and then radiomics features were extracted. A machine learning model was established to evaluate the prediction of PTC aggressiveness.

RESULTS

The study cohort included 107 patients with PTC confirmed by pathology (cross-validation cohort: n = 71; test cohort: n = 36). A total of 1584 features were extracted from contrast-enhanced T1-weighted (CE-T1 WI), T2-weighted (T2 WI) and diffusion weighted (DWI) images of each patient. Sparse representation method is used for radiation feature selection and classification model establishment. The accuracy of the independent test set that using only one modality, like CE-T1WI, T2WI or DWI was not particularly satisfactory. In contrast, the result of these three modalities combined achieved 0.917.

CONCLUSION

Our study shows that multimodality MR image based on radiomics model can accurately distinguish aggressiveness in PTC from non-aggressiveness PTC before operation. This method may be helpful to inform the treatment strategy and prognosis of patients with aggressiveness PTC.

An Ultrasonic-Based Radiomics Nomogram for Distinguishing Between Benign and Malignant Solid Renal Masses

Objectives

This study was conducted in order to develop and validate an ultrasonic-based radiomics nomogram for diagnosing solid renal masses.

Methods

Six hundred renal solid masses with benign renal lesions (n = 204) and malignant renal tumors (n = 396) were divided into a training set (n = 480) and a validation set (n = 120). Radiomics features were extracted from ultrasound (US) images preoperatively and then a radiomics score (RadScore) was calculated. By integrating the RadScore and independent clinical factors, a radiomics nomogram was constructed. The diagnostic performance of junior physician, senior physician, RadScore, and radiomics nomogram in identifying benign from malignant solid renal masses was evaluated based on the area under the receiver operating characteristic curve (ROC) in both the training and validation sets. The clinical usefulness of the nomogram was assessed using decision curve analysis (DCA).

Results

The radiomics signature model showed satisfactory discrimination in the training set [area under the ROC (AUC), 0.887; 95% confidence interval (CI), 0.860–0.915] and the validation set (AUC, 0.874; 95% CI, 0.816–0.932). The radiomics nomogram also demonstrated good calibration and discrimination in the training set (AUC, 0.911; 95% CI, 0.886–0.936) and the validation set (AUC, 0.861; 95% CI, 0.802–0.921). In addition, the radiomics nomogram model showed higher accuracy in discriminating benign and malignant renal masses compared with the evaluations by junior physician (DeLong p = 0.004), and the model also showed significantly higher specificity than the senior and junior physicians (0.93 vs. 0.57 vs. 0.46).

Conclusions

The ultrasonic-based radiomics nomogram shows favorable predictive efficacy in differentiating solid renal masses.

Contrast-Enhanced CT-Based Radiomics for the Differentiation of Nodular Goiter from Papillary Thyroid Carcinoma in Thyroid Nodules

Background

Papillary thyroid carcinoma (PTC) and nodular goiter (NG) represent the most commonly malignant and benign diseases of thyroid nodules and are often confused in diagnosis. CT examination has a certain diagnostic value for the diagnosis of suspected malignant thyroid nodules. The application of machine learning to radiomics features provides a new diagnostic approach, which has been widely used in ultrasound examination of the thyroid, but there are few literatures on CT examination.

Purpose

To explore the efficacy of a diagnostic model aided by machine learning for preoperative differentiation of nodular goiter and papillary thyroid carcinoma thyroid nodules on the basis of 3D arterial-phase contrast-enhanced computed tomography (CECT) features.

Materials and Methods

We collected the data of 193 NG and 214 PTC thyroid nodules from 407 patients in CT examinations. Together with the pathologist findings and radiology diagnosis, we built a radiomics model using the 1218 features extracted from the arterial phase of CECT images. By comparing the diagnostic performance of the radiomics model with that of the clinical diagnosis, we assessed the performance of the radiomics model.

Results

The radiomics model was developed based on multivariable logistic regression with the optimal 12 radiomics features after feature dimension reduction. The radiomics model performed well on the classification accuracy of the PTC and NG thyroid nodules in the training group and validation group.

Conclusion

The radiomics model based on the 3D arterial phase of CECT features performed better than the group of experienced radiologists in differentiating NG and PTC thyroid nodules.

Assessing Diagnostic Value of Combining Ultrasound and MRI in Extrathyroidal Extension of Papillary Thyroid Carcinoma

Purpose

To explore the separate diagnostic value of preoperative ultrasound (US), magnetic resonance imaging (MRI), and the combination of US and MRI in extrathyroidal extension (ETE) of papillary thyroid carcinoma (PTC).

Materials and Methods

This retrospective study was approved by the Affiliated People’s Hospital of Jiangsu University review board. A total of 158 PTC patients with ETE received preoperative US and MRI examination and underwent surgery between May 2014 and December 2018 in Affiliated People’s Hospital of Jiangsu University. For each case, the US and MRI features of ETE were retrospectively and independently investigated by two radiologists. The clinical assessment for each case was implemented, respectively, using US imaging only, MRI only, and a combination of both modalities at three different time points with one-month intervals.

Results

The diagnostic accuracies of US, MRI, and the combined set for T3 (minimal ETE) were 91.7% (88/96), 74.0% (71/96), and 97.9% (94/96), respectively, indicating a significantly different performance (P < 0.001). The diagnostic accuracies for T4 (extensive ETE) were 62.9% (39/62), 87.1% (54/62), and 93.5% (58/62), respectively. The difference between the three methods for T4 was statistically significant (P = 0.000). The diagnostic accuracies for overall ETE were 80.4% (127/158), 79.1% (125/158), and 96.2% (152/158), respectively. The difference between the three methods for ETE was statistically significant (P = 0.001).

Conclusion

This study suggests that ETE can be predicted most accurately by the combination of preoperative US and MRI.

Risk factors of skip lateral cervical lymph node metastasis in papillary thyroid carcinoma: a systematic review and meta-analysis

OBJECTIVE

To analyze and explore the risk factors of skip lateral cervical lymph node metastasis (SLLNM) in papillary thyroid carcinoma (PTC).

METHODS

PubMed, Web of Science, Embase, Cochrane, Wanfang, China National Knowledge Infrastructure, and China Science and Technology Journal databases, updated to April 4, 2021, were systematically searched for literature on the risk factors of SLLNM in PTC. The meta-analysis was completed using Stata 15.0 software after quality evaluation. The odds ratio (OR) and 95% confidence interval (CI) of each variable were calculated using fixed or random-effects models, and the publication bias was evaluated by the Egger's test.

RESULTS

A total of 28 studies with 10,682 cases were included in our meta-analysis; 1592 (14.90%) cases were positive for SLLNM. The meta-analysis showed that female sex (OR = 1.16, 95% CI = 1.02-1.31, P = 0.021), age ≥45 (OR = 1.60, 95% CI = 1.19-2.15, P = 0.002), tumor diameter ≤10 mm (OR = 2.23, 95% CI = 1.62-3.06, P < 0.001), and upper location of tumor (OR = 3.60, 95% CI = 2.65-4.89, P < 0.001) were risk factors for SLLNM in PTC patients. Hashimoto's thyroiditis (OR = 1.02, 95% CI = 0.88-1.19, P = 0.777), multifocality (OR = 0.98, 95% CI = 0.75-1.28, P = 0.873), bilateral tumors (OR = 0.92, 95% CI = 0.70-1.19, P = 0.515), extrathyroidal extensions (OR = 1.07, 95% CI = 0.83-1.39, P = 0.598), and capsular invasion (OR = 0.93, 95% CI = 0.65-1.31, P = 0.660) were not closely related to SLLNM risk.

CONCLUSION

This study confirmed significant associations between SLLNM and female sex, age ≥45, tumor diameter ≤10 mm, and upper location of the tumor.

Application of Machine Learning Methods to Improve the Performance of Ultrasound in Head and Neck Oncology: A Literature Review

Simple Summary

Ultrasound (US) is a non-invasive imaging method that is routinely utilized in head and neck cancer patients to assess the anatomic extent of tumors, nodal and non-nodal neck masses and for imaging the salivary glands. In this review, we summarize the present evidence on whether the application of machine learning (ML) methods can potentially improve the performance of US in head and neck cancer patients. We found that published clinical literature on ML methods applied to US datasets was limited but showed evidence of improved diagnostic and prognostic performance. However, a majority of these studies were based on retrospective evaluation and conducted at a single center with a limited number of datasets. The conduct of multi-center studies could help better validate the performance of ML-based US radiomics and facilitate the integration of these approaches into routine clinical practice.

Abstract

Radiomics is a rapidly growing area of research within radiology that involves the extraction and modeling of high-dimensional quantitative imaging features using machine learning/artificial intelligence (ML/AI) methods. In this review, we describe the published clinical evidence on the application of ML methods to improve the performance of ultrasound (US) in head and neck oncology. A systematic search of electronic databases (MEDLINE, PubMed, clinicaltrials.gov) was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Of 15,080 initial articles identified, 34 studies were selected for in-depth analysis. Twenty-five out of 34 studies (74%) focused on the diagnostic application of US radiomics while 6 (18%) studies focused on response assessment and 3 (8%) studies utilized US radiomics for modeling normal tissue toxicity. Support vector machine (SVM) was the most commonly employed ML method (47%) followed by multivariate logistic regression (24%) and k-nearest neighbor analysis (21%). Only 11/34 (~32%) of the studies included an independent validation set. A majority of studies were retrospective in nature (76%) and based on single-center evaluation (85%) with variable numbers of patients (12–1609) and imaging datasets (32–1624). Despite these limitations, the application of ML methods resulted in improved diagnostic and prognostic performance of US highlighting the potential clinical utility of this approach.

Exploring the Value of Radiomics Features Based on B-Mode and Contrast-Enhanced Ultrasound in Discriminating the Nature of Thyroid Nodules

Background

With the improvement of ultrasound imaging resolution and the application of various new technologies, the detection rate of thyroid nodules has increased greatly in recent years. However, there are still challenges in accurately diagnosing the nature of thyroid nodules. This study aimed to evaluate the clinical application value of the radiomics features extracted from B-mode ultrasound (B-US) images combined with contrast-enhanced ultrasound (CEUS) images in the differentiation of benign and malignant thyroid nodules by comparing the diagnostic performance of four logistic models.

Methods

We retrospectively collected and ultimately included B-US images and CEUS images of 123 nodules from 123 patients, and then extracted the corresponding radiomics features from these images respectively. Meanwhile, a senior radiologist combined the thyroid imaging reporting and data system (TI-RADS) and the enhancement pattern of the ultrasonography to make a graded diagnosis of the malignancy of these nodules. Next, based on these radiomics features and grades, logistic regression was used to help build the models (B-US radiomics model, CEUS radiomics model, B-US+CEUS radiomics model, and TI-RADS+CEUS model). Finally, the study assessed the diagnostic performance of these radiomics features with a comparison of the area under the curve (AUC) of the receiver operating characteristic curve of four logistic models for predicting the benignity or malignancy of thyroid nodules.

Results

The AUC in the differential diagnosis of the nature of thyroid nodules was 0.791 for the B-US radiomics model, 0.766 for the CEUS radiomics model, 0.861 for the B-US+CEUS radiomics model, and 0.785 for the TI-RADS+CEUS model. Compared to the TI-RADS+CEUS model, there was no statistical significance observed in AUC between the B-US radiomics model, CEUS radiomics model, B-US+CEUS radiomics model, and TI-RADS+CEUS model (P>0.05). However, a significant difference was observed between the single B-US radiomics model or CEUS radiomics model and B-US+CEUS radiomics model (P<0.05).

Conclusion

In our study, the B-US radiomics model, CEUS radiomics model, and B-US+CEUS radiomics model demonstrated similar performance with the TI-RADS+CEUS model of senior radiologists in diagnosing the benignity or malignancy of thyroid nodules, while the B-US+CEUS radiomics model showed better diagnostic performance than single B-US radiomics model or CEUS radiomics model. It was proved that B-US radiomics features and CEUS radiomics features are of high clinical value as the combination of the two had better diagnostic performance.

Nomogram Combining Radiomics With the American College of Radiology Thyroid Imaging Reporting and Data System Can Improve Predictive Performance for Malignant Thyroid Nodules

Purpose

To develop and validate a nomogram combining radiomics of B-mode ultrasound (BMUS) images and the American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TI-RADS) for predicting malignant thyroid nodules and improving the performance of the guideline.

Method

A total of 451 thyroid nodules referred for surgery and proven pathologically at an academic referral center from January 2019 to September 2020 were retrospectively collected and randomly assigned to training and validation cohorts (7:3 ratio). A nomogram was developed through combining the BMUS radiomics score (Rad-Score) with ACR TI-RADS score (ACR-Score) in the training cohort; the performance of the nomogram was assessed with respect to discrimination, calibration, and clinical application in the validation and entire cohorts.

Results

The ACR-Rad nomogram showed good calibration and yielded an AUC of 0.877 (95% CI 0.836–0.919) in the training cohort and 0.864 (95% CI 0.799–0.931) in the validation cohort, which were significantly better than the ACR-Score model (p < 0.001 and 0.031, respectively). The significantly improved AUC, net reclassification index (NRI), and integrated discriminatory improvement (IDI) of the nomogram were found for both senior and junior radiologists (all p < 0.001). Decision curve analysis indicated that the nomogram was clinically useful. When cutoff values for 50% predicted malignancy risk (ACR-Rad_50%) were applied, the nomogram showed increased specificity, accuracy and positive predictive value (PPV), and decreased unnecessary fine-needle aspiration (FNA) rates in comparison to ACR TI-RADS.

Conclusion

The ACR-Rad nomogram has favorable value in predicting malignant thyroid nodules and improving performance of the ACR TI-RADS for senior and junior radiologists.

Diagnostic Utility of Radiomics in Thyroid and Head and Neck Cancers

Radiomics is an emerging field in radiology that utilizes advanced statistical data characterizing algorithms to evaluate medical imaging and objectively quantify characteristics of a given disease. Due to morphologic heterogeneity and genetic variation intrinsic to neoplasms, radiomics have the potential to provide a unique insight into the underlying tumor and tumor microenvironment. Radiomics has been gaining popularity due to potential applications in disease quantification, predictive modeling, treatment planning, and response assessment - paving way for the advancement of personalized medicine. However, producing a reliable radiomic model requires careful evaluation and construction to be translated into clinical practices that have varying software and/or medical equipment. We aim to review the diagnostic utility of radiomics in otorhinolaryngology, including both cancers of the head and neck as well as the thyroid.

Radiomics Nomogram for Identifying Sub-1 cm Benign and Malignant Thyroid Lesions

PURPOSE

To develop and validate a radiomics nomogram for identifying sub-1 cm benign and malignant thyroid lesions.

METHOD

A total of 171 eligible patients with sub-1 cm thyroid lesions (56 benign and 115 malignant) who were treated in Yantai Yuhuangding Hospital between January and September 2019 were retrospectively collected and randomly divided into training (n = 136) and validation sets (n = 35). The radiomics features were extracted from unenhanced and arterial contrast-enhanced computed tomography images of each patient. In the training set, one-way analysis of variance and least absolute shrinkage and selection operator (LASSO) logistic regression were used to select the features related to benign and malignant lesions, and the LASSO algorithm was used to construct the radiomics signature. Combined with clinical independent predictive factors, a radiomics nomogram was constructed with a multivariate logistic regression model. The performance of the radiomics nomogram was evaluated by using the receiver operating characteristic (ROC) and calibration curves in the training and validation sets. The clinical usefulness was evaluated by using decision curve analysis (DCA).

RESULTS

The radiomics signature consisting of 13 selected features achieved favorable prediction efficiency. The radiomics nomogram, which incorporated radiomics signature and clinical independent predictive factors including age and Thyroid Imaging Reporting and Data System category, showed good calibration and discrimination in the training (area under the ROC [AUC]: 0.853; 95% confidence interval [CI]: 0.797, 0.899) and validation sets (AUC: 0.851; 95% CI: 0.735, 0.931). DCA demonstrated that the nomogram was clinically useful.

CONCLUSION

As a noninvasive preoperative prediction tool, the radiomics nomogram incorporating radiomics signature and clinical predictive factors shows favorable predictive efficiency for identifying sub-1 cm benign and malignant thyroid lesions.

Radiomics Score Combined with ACR TI-RADS in Discriminating Benign and Malignant Thyroid Nodules Based on Ultrasound Images: A Retrospective Study

This study aimed to explore the ability of combination model of ultrasound radiomics score (Rad-score) and the thyroid imaging, reporting and data system by the American College of Radiology (ACR TI-RADS) in predicting benign and malignant thyroid nodules (TNs). Up to 286 radiomics features were extracted from ultrasound images of TNs. By using the lowest probability of classification error and average correlation coefficients (POE + ACC) and the least absolute shrinkage and selection operator (LASSO), we finally selected four features to establish Rad-score (Vertl-RLNonUni, Vertl-GLevNonU, WavEnLH-s4 and WavEnHL-s5). DeLong’s test and decision curve analysis (DCA) showed that the method of combining Rad-score and ACR TI-RADS had the best performance (the area under the receiver operating characteristic curve (AUC = 0.913 (95% confidence interval (CI), 0.881–0.939) and 0.899 (95%CI, 0.840–0.942) in the training group and verification group, respectively), followed by ACR TI-RADS (AUC = 0.898 (95%CI, 0.863–0.926) and 0.870 (95%CI, 0.806–0.919) in the training group and verification group, respectively), and followed by Rad-score (AUC = 0.750 (95%CI, 0.704–0.792) and 0.750 (95%CI, 0.672–0.817) in the training group and verification group, respectively). We concluded that the ability of ultrasound Rad-score to distinguish benign and malignant TNs was not as good as that of ACR TI-RADS, and the ability of the combination model of Rad-score and ACR TI-RADS to discriminate benign and malignant TNs was better than ACR TI-RADS or Rad-score alone. Ultrasound Rad-score might play a potential role in improving the differentiation of malignant TNs from benign TNs.

Deep learning-based artificial intelligence model to assist thyroid nodule diagnosis and management: a multicentre diagnostic study

BACKGROUND

Strategies for integrating artificial intelligence (AI) into thyroid nodule management require additional development and testing. We developed a deep-learning AI model (ThyNet) to differentiate between malignant tumours and benign thyroid nodules and aimed to investigate how ThyNet could help radiologists improve diagnostic performance and avoid unnecessary fine needle aspiration.

METHODS

ThyNet was developed and trained on 18 049 images of 8339 patients (training set) from two hospitals (the First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China, and Sun Yat-sen University Cancer Center, Guangzhou, China) and tested on 4305 images of 2775 patients (total test set) from seven hospitals (the First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangzhou, China; the Sixth Affiliated Hospital of Sun Yat-sen University, Guangzhou, China; the Guangzhou Army General Hospital, Guangzhou, China; the Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China; the First Affiliated Hospital of Sun Yat-sen University; Sun Yat-sen University Cancer Center; and the First Affiliated Hospital of Guangxi Medical University, Nanning, China) in three stages. All nodules in the training and total test set were pathologically confirmed. The diagnostic performance of ThyNet was first compared with 12 radiologists (test set A); a ThyNet-assisted strategy, in which ThyNet assisted diagnoses made by radiologists, was developed to improve diagnostic performance of radiologists using images (test set B); the ThyNet assisted strategy was then tested in a real-world clinical setting (using images and videos; test set C). In a simulated scenario, the number of unnecessary fine needle aspirations avoided by ThyNet-assisted strategy was calculated.

FINDINGS

The area under the receiver operating characteristic curve (AUROC) for accurate diagnosis of ThyNet (0·922 [95% CI 0·910-0·934]) was significantly higher than that of the radiologists (0·839 [0·834-0·844]; p<0·0001). Furthermore, ThyNet-assisted strategy improved the pooled AUROC of the radiologists from 0·837 (0·832-0·842) when diagnosing without ThyNet to 0·875 (0·871-0·880; p<0·0001) with ThyNet for reviewing images, and from 0·862 (0·851-0·872) to 0·873 (0·863-0·883; p<0·0001) in the clinical test, which used images and videos. In the simulated scenario, the number of fine needle aspirations decreased from 61·9% to 35·2% using the ThyNet-assisted strategy, while missed malignancy decreased from 18·9% to 17·0%.

INTERPRETATION

The ThyNet-assisted strategy can significantly improve the diagnostic performance of radiologists and help reduce unnecessary fine needle aspirations for thyroid nodules.

FUNDING

National Natural Science Foundation of China and Guangzhou Science and Technology Project.

A Comparative Analysis of Two Machine Learning-Based Diagnostic Patterns with Thyroid Imaging Reporting and Data System for Thyroid Nodules: Diagnostic Performance and Unnecessary Biopsy Rate

Background: The risk stratification system of the American College of Radiology Thyroid Imaging Reporting and Data System (ACR TI-RADS) for thyroid nodules is affected by low diagnostic specificity. Machine learning (ML) methods can optimize the diagnostic performance in medical image analysis. However, it is unknown which ML-based diagnostic pattern is more effective in improving diagnostic performance for thyroid nodules and reducing nodule biopsies. Therefore, we compared ML-assisted visual approaches and radiomics approaches with ACR TI-RADS in diagnostic performance and unnecessary fine-needle aspiration biopsy (FNAB) rate for thyroid nodules. Methods: This retrospective study evaluated a data set of ultrasound (US) and shear wave elastography (SWE) images in patients with biopsy-proven thyroid nodules (≥1 cm) from the Shanghai Tenth People's Hospital (743 nodules in 720 patients from September 2017 to January 2019) and an independent test data set from the Ma'anshan People's Hospital (106 nodules in 102 patients from February 2019 to April 2019). Six US features and five SWE parameters from the radiologists' interpretation were used for building the ML-assisted visual approaches. The radiomics features extracted from the US and SWE images were used with ML methods for developing the radiomics approaches. The diagnostic performance for differentiating thyroid nodules and the unnecessary FNAB rate of the ML-assisted visual approaches and the radiomics approaches were compared with ACR TI-RADS. Results: The ML-assisted US visual approach had the best diagnostic performance than the US radiomics approach and ACR TI-RADS (area under the curve [AUC]: 0.900 vs. 0.789 vs. 0.689 for the validation data set, 0.917 vs. 0.770 vs. 0.681 for the test data set). After adding SWE, the ML-assisted visual approach had a better diagnostic performance than US alone (AUC: 0.951 vs. 0.900 for the validation data set, 0.953 vs. 0.917 for the test data set). When applying the ML-assisted US+SWE visual approach, the unnecessary FNAB rate decreased from 30.0% to 4.5% in the validation data set and from 37.7% to 4.7% in the test data set in comparison to ACR TI-RADS. Conclusions: The ML-assisted dual modalities visual approach can assist radiologists to diagnose thyroid nodules more effectively and considerably reduce the unnecessary FNAB rate in the clinical management of thyroid nodules.

Radiomics based on multiparametric MRI for extrathyroidal extension feature prediction in papillary thyroid cancer

BACKGROUND

To determine the predictive capability of MRI-based radiomics for extrathyroidal extension detection in papillary thyroid cancer (PTC) pre-surgically.

METHODS

The present retrospective trial assessed individuals with thyroid nodules examined by multiparametric MRI and subsequently administered thyroid surgery. Diagnosis and extrathyroidal extension (ETE) feature of PTC were based on pathological assessment. The thyroid tumors underwent manual segmentation, for radiomic feature extraction. Participants were randomized to the training and testing cohorts, at a ratio of 7:3. The mRMR (maximum correlation minimum redundancy) algorithm and the least absolute shrinkage and selection operator were utilized for radiomics feature selection. Then, a radiomics predictive model was generated via a linear combination of the features. The model's performance in distinguishing the ETE feature of PTC was assessed by analyzing the receiver operating characteristic curve.

RESULTS

Totally 132 patients were assessed in this study, including 92 and 40 in the training and test cohorts, respectively). Next, the 16 top-performing features, including 4, 7 and 5 from diffusion weighted (DWI), T2-weighted (T2 WI), and contrast-enhanced T1-weighted (CE-T1WI) images, respectively, were finally retained to construct the radiomics signature. There were 8 RLM, 5 CM, 2 shape, and 1 SZM features. The radiomics prediction model achieved AUCs of 0.96 and 0.87 in the training and testing sets, respectively.

CONCLUSIONS

Our study indicated that MRI radiomics approach had the potential to stratify patients based on ETE in PTCs preoperatively.

Implications of US radiomics signature for predicting malignancy in thyroid nodules with indeterminate cytology

OBJECTIVES

The purpose of this study was to evaluate the role of the radiomics score using US images to predict malignancy in AUS/FLUS and FN/SFN nodules.

METHODS

One hundred fifty-five indeterminate thyroid nodules in 154 patients who received initial US-guided FNA for diagnostic purposes were included in this retrospective study. A representative US image of each tumor was acquired, and square ROIs covering the whole nodule were drawn using the Paint program of Windows 7. Texture features were extracted by in-house texture analysis algorithms implemented in MATLAB 2019b. The LASSO logistic regression model was used to choose the most useful predictive features, and ten-fold cross-validation was performed. Two prediction models were constructed using multivariable logistic regression analysis: one based on clinical variables, and the other based on clinical variables with the radiomics score. Predictability of the two models was assessed with the AUC of the ROC curves.

RESULTS

Clinical characteristics did not significantly differ between malignant and benign nodules, except for mean nodule size. Among 730 candidate texture features generated from a single US image, 15 features were selected. Radiomics signatures were constructed with a radiomics score, using selected features. In multivariable logistic regression analysis, higher radiomics score was associated with malignancy (OR = 10.923; p < 0.001). The AUC of the malignancy prediction model composed of clinical variables with the radiomics score was significantly higher than the model composed of clinical variables alone (0.839 vs 0.583).

CONCLUSIONS

Quantitative US radiomics features can help predict malignancy in thyroid nodules with indeterminate cytology.

Shear wave elastography-based ultrasomics: differentiating malignant from benign focal liver lesions

PURPOSE

Ultrasomics is a radiomics technique that extracts high-throughput quantitative data from ultrasound imaging. The aim of this study was to differentiate malignant from benign focal liver lesions (FLLs) using two-dimensional shear wave elastography (2D-SWE)-based ultrasomics.

METHODS

A total of 175 FLLs in 169 patients were prospectively analyzed. The study population was divided into a training cohort (n = 122) and a validation cohort (n = 53). The maxima, minima, mean, and standard deviation of 2D-SWE measurements were expressed in kilopascals (E_max, E_min, E_mean, and E_SD). The ultrasonics technique was used to extract the features from the 2D-SWE images. Support vector machine was used to establish two prediction models: the ultrasomics score (ultrasomics features only) and the combined score (SWE measurements and ultrasomics features). The diagnostic performance of the models in differentiating FLLs was analyzed.

RESULTS

A total of 1044 features were extracted and 15 features were selected. The AUC for the combined score, ultrasomics score, E_max, E_mean, E_min and E_SD were 0.94, 0.91, 0.92, 0.89, 0.67, and 0.89, respectively. The combined score had the best diagnostic performance. The sensitivity, specificity, PPV, NPV, +LR, LR of the combined score were 92.59%, 87.50%, 94.59%, 82.50%, 7.35%, and 0.09%, respectively. The decision curve analysis results showed that when the threshold probability was > 29%, the combined score showed improved benefits for patients compared to using the ultrasomics score and 2D-SWE measurements.

CONCLUSION

The results of this study demonstrated that the combined score had good diagnostic accuracy in differentiating malignant from benign FLLs.

Longitudinal radiomics algorithm of posttreatment computed tomography images for early detecting recurrence of hepatocellular carcinoma after resection or ablation

OBJECTIVES

To develop a radiomics algorithm, improving the performance of detecting recurrence, based on posttreatment CT images within one month and at suspicious time during follow-up.

MATERIALS AND METHODS

A total of 114 patients with 228 images were randomly split (7:3) into training and validation cohort. Radiomics algorithm was trained using machine learning, based on difference-in-difference (DD) features extracted from tumor and liver regions of interest on posttreatment CTs within one month after resection or ablation and when suspected recurrent lesion was observed but cannot be confirmed as HCC during follow-up. The performance was evaluated by area under the receiver operating characteristic curve (AUC) and was compared among radiomics algorithm, change of alpha-fetoprotein (AFP) and combined model of both. Five-folded cross validation (CV) was used to present the training error.

RESULTS

A radiomics algorithm was established by 34 DD features selected by random forest and multivariable logistic models and showed a better AUC than that of change of AFP (0.89 [95% CI: 0.78, 1.00] vs 0.63 [95% CI: 0.42, 0.84], P = .04) and similar with the combined model in detecting recurrence in the validation set. Five-folded CV error in the validation cohort was 21% for the algorithm and 26% for the changes of AFP.

CONCLUSIONS

The algorithm integrated radiomic features of posttreatment CT showed superior performance to that of conventional AFP and may act as a potential marker in the early detecting recurrence of HCC.

Combining radiomics with ultrasound-based risk stratification systems for thyroid nodules: an approach for improving performance

OBJECTIVES

To develop a radiomics score using ultrasound images to predict thyroid malignancy and to investigate its potential as a complementary tool to improve the performance of risk stratification systems.

METHODS

We retrospectively included consecutive patients who underwent fine-needle aspiration (FNA) for thyroid nodules that were cytopathologically diagnosed as benign or malignant. Nodules were randomly assigned to a training and test set (8:2 ratio). A radiomics score was developed from the training set, and cutoff values based on the maximum Youden index (Rad_maxY) and for 5%, 10%, and 20% predicted malignancy risk (Rad_5%, Rad_10%, Rad_20%, respectively) were applied to the test set. The performances of the American College of Radiology (ACR) and the American Thyroid Association (ATA) guidelines were compared with the combined performances of the guidelines and radiomics score with interpretations from expert and nonexpert readers.

RESULTS

A total of 1624 thyroid nodules from 1609 patients (mean age, 50.1 years [range, 18-90 years]) were included. The radiomics score yielded an AUC of 0.85 (95% CI: 0.83, 0.87) in the training set and 0.75 (95% CI: 0.69, 0.81) in the test set (Rad_maxY). When the radiomics score was combined with the ACR or ATA guidelines (Rad_5%), all readers showed increased specificity, accuracy, and PPV and decreased unnecessary FNA rates (all p < .05), with no difference in sensitivity (p > .05).

CONCLUSION

Radiomics help predict thyroid malignancy and improve specificity, accuracy, PPV, and unnecessary FNA rate while maintaining the sensitivity of the ACR and ATA guidelines for both expert and nonexpert readers.

KEY POINTS

• The radiomics score yielded an AUC of 0.85 and 0.75 in the training and test set, respectively. • For all readers, combining a 5% predicted malignancy risk cutoff for the radiomics score with the ACR and ATA guidelines significantly increased specificity, accuracy, and PPV and decreased unnecessary FNA rates, with no decrease in sensitivity. • Radiomics can help predict malignancy in thyroid nodules in combination with risk stratification systems, by improving specificity, accuracy, and PPV and unnecessary FNA rates while maintaining sensitivity for both expert and nonexpert readers.