Partially cystic thyroid cancer on conventional and elastographic ultrasound: a retrospective study and a machine learning—assisted system

Zhao, Hanqing; Liu, Jingyan; Lin, Qingyu; He, Yaqian; Luo, Honghao; Peng, Yulan; Ma, Buyun

doi:10.21037/atm.2020.03.211

Cited by 13 publications

(9 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In univariate analysis, we did not find that nodule size was significantly associated with malignancy of PCTNs, and the finding was in accordance with previous studies (11,19,30). Hence, nodule size has not been described as a risk factor of malignant PCTNs.…”

Section: Figuresupporting

confidence: 88%

A Scoring System for Assessing the Risk of Malignant Partially Cystic Thyroid Nodules Based on Ultrasound Features

et al. 2021

View full text Add to dashboard Cite

ObjectiveTo assess the ultrasound (US) features of partially cystic thyroid nodules (PCTNs) and to establish a scoring system to further improve the diagnostic accuracy.MethodsA total of 262 consecutive nodules from September 2017 to March 2020 were included in a primary cohort to construct a scoring system. Moreover, 83 consecutive nodules were enrolled as an validation cohort from May 2018 to August 2020. All nodules were determined to be benign or malignant according to the pathological results after surgery or ultrasound-guided fine-needle aspiration (US-FNA). The US images and demographic characteristics of the patients were analyzed. The ultrasound features of PCTNs were extracted from primary cohort by two experienced radiologists. The features extracted were used to develop a scoring system using logistic regression analysis. Receiver operating characteristic (ROC) curves were applied to evaluate the diagnostic efficacy of the scoring system in both the primary cohort and validation cohort. In addition, the radiologists evaluated the benign and malignant PCTNs of the validation cohort according to the ACR TI-RADS guidelines and clinical experience, and the accuracy of their diagnosis were compared with that of the scoring system.ResultsBased on the eight features of PCTNs, the scoring system showed good differentiation and reproducibility in both cohorts. The scoring system was based on eight features of PCTNs and showed good performance. The area under the curve (AUC) was 0.876 (95% CI, 0.830 - 0.913) in the primary cohort and 0.829(95% CI, 0.730 - 0.903) in the validation cohort. The optimal cutoff value of the scoring system for the diagnosis of malignant PCTNs was 4 points, with a good sensitivity of 71.05% and specificity of 87.63%. The scoring system (AUC=0.829) was superior to radiologists (AUC= 0.736) in diagnosing PCTNs and is a promising method for clinical application.ConclusionsThe scoring system described herein is a convenient and clinically valuable method that can diagnose PCTNs with relatively high accuracy. The use of this method to diagnose PCTNs, which have been previously underestimated, will allow PCTNs to receive reasonable attention, and assist radiologist to confidently diagnose the benignity or malignancy.

show abstract

Section: Figuresupporting

confidence: 88%

A Scoring System for Assessing the Risk of Malignant Partially Cystic Thyroid Nodules Based on Ultrasound Features

et al. 2021

View full text Add to dashboard Cite

show abstract

“…In fact, we found that a massive number of patients harbor clinically significant lymph node metastases which have not been detected by pre-oprative ultrasound. Furthermore, while ML approaches have shown unparalleled diagnostic performance in differentiating between benign and malignant thyroid nodules in recent reports ( 35 , 36 ), there is, however, little research in the available literature on applying ML algorithms to lymph node metastases in PTC. To the best of our knowledge, this is the very first study to develop a prediction model using ML algorithms for real-time risk evaluation of CLNM with easy-to-use clinical data and fortunately, our model shows a great predictive power, which distinguishes itself from linear models adopted by previous researches.…”

Section: Discussionmentioning

confidence: 99%

Application of Machine Learning Algorithms to Predict Central Lymph Node Metastasis in T1-T2, Non-invasive, and Clinically Node Negative Papillary Thyroid Carcinoma

Zhu

Zheng

et al. 2021

Front. Med.

View full text Add to dashboard Cite

Purpose: While there are no clear indications of whether central lymph node dissection is necessary in patients with T1-T2, non-invasive, clinically uninvolved central neck lymph nodes papillary thyroid carcinoma (PTC), this study seeks to develop and validate models for predicting the risk of central lymph node metastasis (CLNM) in these patients based on machine learning algorithms.Methods: This is a retrospective study comprising 1,271 patients with T1-T2 stage, non-invasive, and clinically node negative (cN0) PTC who underwent surgery at the Department of Endocrine and Breast Surgery of The First Affiliated Hospital of Chongqing Medical University from February 1, 2016, to December 31, 2018. We applied six machine learning (ML) algorithms, including Logistic Regression (LR), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGBoost), Random Forest (RF), Decision Tree (DT), and Neural Network (NNET), coupled with preoperative clinical characteristics and intraoperative information to develop prediction models for CLNM. Among all the samples, 70% were randomly selected to train the models while the remaining 30% were used for validation. Indices like the area under the receiver operating characteristic (AUROC), sensitivity, specificity, and accuracy were calculated to test the models' performance.Results: The results showed that ~51.3% (652 out of 1,271) of the patients had pN1 disease. In multivariate logistic regression analyses, gender, tumor size and location, multifocality, age, and Delphian lymph node status were all independent predictors of CLNM. In predicting CLNM, six ML algorithms posted AUROC of 0.70–0.75, with the extreme gradient boosting (XGBoost) model standing out, registering 0.75. Thus, we employed the best-performing ML algorithm model and uploaded the results to a self-made online risk calculator to estimate an individual's probability of CLNM (https://jin63.shinyapps.io/ML_CLNM/).Conclusions: With the incorporation of preoperative and intraoperative risk factors, ML algorithms can achieve acceptable prediction of CLNM with Xgboost model performing the best. Our online risk calculator based on ML algorithm may help determine the optimal extent of initial surgical treatment for patients with T1-T2 stage, non-invasive, and clinically node negative PTC.

show abstract

“…Without a threshold effect, we would directly calculate the pooled sensitivity (Se), specificity (Sp), positive and negative likelihood ratios (LR+ and LR−), diagnostic OR (DOR), and area under the curve (AUC). A hierarchical summary receiver operating characteristic curve (HSROC) should be used to calculate AUC when a threshold exists (26)(27)(28)(29). Table 1 demonstrates the basic information of the eight included studies.…”

Section: Discussionmentioning

confidence: 99%

“…Some features are closely associated and can exist simultaneously in malignant nodules ( 40 ); however the inherent relationship between suspicious US features could not be explored and we failed to evaluate the diagnostic value of combined US features. Then, more detailed classification of specific US feature could bring new insight, but we failed to do such research: for instance, included studies ( 19 , 20 , 29 ) in our review did not divide the degree of hypoechogenicity when exploring associated factors for malignancy, which limited our advanced analysis. Further study could be conducted to find the relationship between degree of hypoechogenicity and malignancy.…”

Section: Discussionmentioning

confidence: 99%

Diagnostic Value of Sonographic Features in Distinguishing Malignant Partially Cystic Thyroid Nodules: A Systematic Review and Meta-Analysis

et al. 2021

View full text Add to dashboard Cite

Ultrasonography (US) is one of the most important methods for the management of thyroid nodules, which can be classified as solid, partially cystic, or cystic by composition. The various Thyroid Imaging Reporting and Data System classifications pay more attention to solid nodules and have reported pertinent US features associated with malignancy. However, the likelihood of malignancy of partially cystic thyroid nodules (PCTNs) is 3.3–17.6%, and few studies have systematically discussed the value of US in differentiating such entities. Therefore, we deemed it necessary to perform a systematic evaluation of US features in recognizing malignant PCTNs. Our systematic review and meta-analysis aimed to assess the value of US features in predicting malignant PCTNs. We searched the PubMed/MEDLINE, Web of Science, and Cochrane Library databases to find studies that researched US features of PCTNs and that were published before June 2020. Review Manager 5.3 was used to summarize suspicious US features and calculate the sensitivity, specificity, and likelihood ratios. MetaDiSc 1.4 was used to estimate receiver operating characteristic curves and calculate areas under the curves (AUCs). Our review included eight studies with a total of 2,004 PCTNs. Seven features were considered to be associated with malignancy. High specificity (>0.9) was found in nodules with a taller-than-wide shape, those that were spiculated/microlobulated or with an ill-defined margin, those with microcalcification, and a non-smooth rim. Among US features, eccentric configuration, microcalcification, and marked or mild hypoechogenicity were more reliable in predicting malignancy (AUC: 0.9592, 0.8504, and 0.8092, respectively). After meta-analysis, we recommend combining PCTN US features including an eccentric internal solid portion, marked or mild hypoechogenicity, and presence of microcalcification to better identify malignant nodules. More studies are needed to explore and improve the diagnostic value of US in PCTNs.

show abstract

Partially cystic thyroid cancer on conventional and elastographic ultrasound: a retrospective study and a machine learning—assisted system

Cited by 13 publications

References 24 publications

A Scoring System for Assessing the Risk of Malignant Partially Cystic Thyroid Nodules Based on Ultrasound Features

A Scoring System for Assessing the Risk of Malignant Partially Cystic Thyroid Nodules Based on Ultrasound Features

Application of Machine Learning Algorithms to Predict Central Lymph Node Metastasis in T1-T2, Non-invasive, and Clinically Node Negative Papillary Thyroid Carcinoma

Diagnostic Value of Sonographic Features in Distinguishing Malignant Partially Cystic Thyroid Nodules: A Systematic Review and Meta-Analysis

Contact Info

Product

Resources

About