Differentiation of thyroid nodules on US using features learned and extracted from various convolutional neural networks

Ultrasound-based nomogram to predict the recurrence in papillary thyroid carcinoma using machine learning

BACKGROUND AND AIMS

The recurrence of papillary thyroid carcinoma (PTC) is not unusual and associated with risk of death. This study is aimed to construct a nomogram that combines clinicopathological characteristics and ultrasound radiomics signatures to predict the recurrence in PTC.

METHODS

A total of 554 patients with PTC who underwent ultrasound imaging before total thyroidectomy were included. Among them, 79 experienced at least one recurrence. Then 388 were divided into the training cohort and 166 into the validation cohort. The radiomics features were extracted from the region of interest (ROI) we manually drew on the tumor image. The feature selection was conducted using Cox regression and least absolute shrinkage and selection operator (LASSO) analysis. And multivariate Cox regression analysis was used to build the combined nomogram using radiomics signatures and significant clinicopathological characteristics. The efficiency of the nomogram was evaluated by receiver operating characteristic (ROC) curves, calibration curves and decision curve analysis (DCA). Kaplan-Meier analysis was used to analyze the recurrence-free survival (RFS) in different radiomics scores (Rad-scores) and risk scores.

RESULTS

The combined nomogram demonstrated the best performance and achieved an area under the curve (AUC) of 0.851 (95% CI: 0.788 to 0.913) in comparison to that of the radiomics signature and the clinical model in the training cohort at 3 years. In the validation cohort, the combined nomogram (AUC = 0.885, 95% CI: 0.805 to 0.930) also performed better. The calibration curves and DCA verified the clinical usefulness of combined nomogram. And the Kaplan-Meier analysis showed that in the training cohort, the cumulative RFS in patients with higher Rad-score was significantly lower than that in patients with lower Rad-score (92.0% vs. 71.9%, log rank P < 0.001), and the cumulative RFS in patients with higher risk score was significantly lower than that in patients with lower risk score (97.5% vs. 73.5%, log rank P < 0.001). In the validation cohort, patients with a higher Rad-score and a higher risk score also had a significantly lower RFS.

CONCLUSION

We proposed a nomogram combining clinicopathological variables and ultrasound radiomics signatures with excellent performance for recurrence prediction in PTC patients.

Value of deep learning models based on ultrasonic dynamic videos for distinguishing thyroid nodules

Objective

This study was designed to distinguish benign and malignant thyroid nodules by using deep learning(DL) models based on ultrasound dynamic videos.

Methods

Ultrasound dynamic videos of 1018 thyroid nodules were retrospectively collected from 657 patients in Zhejiang Cancer Hospital from January 2020 to December 2020 for the tests with 5 DL models.

Results

In the internal test set, the area under the receiver operating characteristic curve (AUROC) was 0.929(95% CI: 0.888,0.970) for the best-performing model LSTM Two radiologists interpreted the dynamic video with AUROC values of 0.760 (95% CI: 0.653, 0.867) and 0.815 (95% CI: 0.778, 0.853). In the external test set, the best-performing DL model had AUROC values of 0.896(95% CI: 0.847,0.945), and two ultrasound radiologist had AUROC values of 0.754 (95% CI: 0.649,0.850) and 0.833 (95% CI: 0.797,0.869).

Conclusion

This study demonstrates that the DL model based on ultrasound dynamic videos performs better than the ultrasound radiologists in distinguishing thyroid nodules.

A comparison of artificial intelligence versus radiologists in the diagnosis of thyroid nodules using ultrasonography: a systematic review and meta-analysis

BACKGROUND

Thyroid nodules are common. Ultrasonography (US) is the first investigation for thyroid nodules. Artificial Intelligence (AI) is widely integrated into medical diagnosis to provide additional information. The primary objective of this study was to accumulate the pooled sensitivity and specificity between all available AI and radiologists using thyroid US imaging. The secondary objective was to compare AI's diagnostic performance to that of radiologists.

MATERIALS AND METHODS

A systematic review meta-analysis. PubMed, Scopus, Web of Science, and Cochrane Library data were searched for studies from inception until June 11, 2020.

RESULTS

Twenty five studies were included in this meta-analysis. The pooled sensitivity and specificity of AI were 0.86 (95% CI 0.81-0.91) and 0.78 (95% CI 0.73-0.83), respectively. The pooled sensitivity and specificity of radiologists were 0.85 (95% CI 0.80-0.89) and 0.82 (95% CI 0.77-0.86), respectively. The accuracy of AI and radiologists is equivalent in terms of AUC [AI 0.89 (95% CI 0.86-0.92), radiologist 0.91 (95% CI 0.88-0.93)]. The diagnostic odd ratio (DOR) between AI 23.10 (95% CI 14.20-37.58) and radiologists 27.12 (95% CI 17.45-42.16) had no statistically significant difference (P = 0.56). Meta-regression analysis revealed that Deep Learning AI had significantly greater sensitivity and specificity than classic machine learning AI (P < 0.001).

CONCLUSION

AI demonstrated comparable performance to radiologists in diagnosing benign and malignant thyroid nodules using ultrasonography. Additional research to establish its equivalency should be conducted.

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.

Machine Learning: Applications and Advanced Progresses of Radiomics in Endocrine Neoplasms

Endocrine neoplasms remain a great threat to human health. It is extremely important to make a clear diagnosis and timely treatment of endocrine tumors. Machine learning includes radiomics, which has long been utilized in clinical cancer research. Radiomics refers to the extraction of valuable information by analyzing a large amount of standard data with high-throughput medical images mainly including computed tomography, positron emission tomography, magnetic resonance imaging, and ultrasound. With the quantitative imaging analysis and model building, radiomics can reflect specific underlying characteristics of a disease that otherwise could not be evaluated visually. More and more promising results of radiomics in oncological practice have been seen in recent years. Radiomics may have the potential to supplement traditional imaging analysis and assist in providing precision medicine for patients. Radiomics had developed rapidly in endocrine neoplasms practice in the past decade. In this review, we would introduce the general workflow of radiomics and summarize the applications and developments of radiomics in endocrine neoplasms in recent years. The limitations of current radiomic research studies and future development directions would also be discussed.

Diagnosing thyroid nodules with atypia of undetermined significance/follicular lesion of undetermined significance cytology with the deep convolutional neural network

To compare the diagnostic performances of physicians and a deep convolutional neural network (CNN) predicting malignancy with ultrasonography images of thyroid nodules with atypia of undetermined significance (AUS)/follicular lesion of undetermined significance (FLUS) results on fine-needle aspiration (FNA). This study included 202 patients with 202 nodules ≥ 1 cm AUS/FLUS on FNA, and underwent surgery in one of 3 different institutions. Diagnostic performances were compared between 8 physicians (4 radiologists, 4 endocrinologists) with varying experience levels and CNN, and AUS/FLUS subgroups were analyzed. Interobserver variability was assessed among the 8 physicians. Of the 202 nodules, 158 were AUS, and 44 were FLUS; 86 were benign, and 116 were malignant. The area under the curves (AUCs) of the 8 physicians and CNN were 0.680–0.722 and 0.666, without significant differences (P > 0.05). In the subgroup analysis, the AUCs for the 8 physicians and CNN were 0.657–0.768 and 0.652 for AUS, 0.469–0.674 and 0.622 for FLUS. Interobserver agreements were moderate (k = 0.543), substantial (k = 0.652), and moderate (k = 0.455) among the 8 physicians, 4 radiologists, and 4 endocrinologists. For thyroid nodules with AUS/FLUS cytology, the diagnostic performance of CNN to differentiate malignancy with US images was comparable to that of physicians with variable experience levels.

Convolutional Neural Network to Stratify the Malignancy Risk of Thyroid Nodules: Diagnostic Performance Compared with the American College of Radiology Thyroid Imaging Reporting and Data System Implemented by Experienced Radiologists

BACKGROUND AND PURPOSE

Comparison of the diagnostic performance for thyroid cancer on ultrasound between a convolutional neural network and visual assessment by radiologists has been inconsistent. Thus, we aimed to evaluate the diagnostic performance of the convolutional neural network compared with the American College of Radiology Thyroid Imaging Reporting and Data System (TI-RADS) for the diagnosis of thyroid cancer using ultrasound images.

MATERIALS AND METHODS

From March 2019 to September 2019, seven hundred sixty thyroid nodules (≥10 mm) in 757 patients were diagnosed as benign or malignant through fine-needle aspiration, core needle biopsy, or an operation. Experienced radiologists assessed the sonographic descriptors of the nodules, and 1 of 5 American College of Radiology TI-RADS categories was assigned. The convolutional neural network provided malignancy risk percentages for nodules based on sonographic images. Sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were calculated with cutoff values using the Youden index and compared between the convolutional neural network and the American College of Radiology TI-RADS. Areas under the receiver operating characteristic curve were also compared.

RESULTS

Of 760 nodules, 176 (23.2%) were malignant. At an optimal threshold derived from the Youden index, sensitivity and negative predictive values were higher with the convolutional neural network than with the American College of Radiology TI-RADS (81.8% versus 73.9%, P = .009; 94.0% versus 92.2%, P = .046). Specificity, accuracy, and positive predictive values were lower with the convolutional neural network than with the American College of Radiology TI-RADS (86.1% versus 93.7%, P < .001; 85.1% versus 89.1%, P = .003; and 64.0% versus 77.8%, P < .001). The area under the curve of the convolutional neural network was higher than that of the American College of Radiology TI-RADS (0.917 versus 0.891, P = .017).

CONCLUSIONS

The convolutional neural network provided diagnostic performance comparable with that of the American College of Radiology TI-RADS categories assigned by experienced radiologists.

Artificial intelligence in ultrasound

Ultrasound (US), a flexible green imaging modality, is expanding globally as a first-line imaging technique in various clinical fields following with the continual emergence of advanced ultrasonic technologies and the well-established US-based digital health system. Actually, in US practice, qualified physicians should manually collect and visually evaluate images for the detection, identification and monitoring of diseases. The diagnostic performance is inevitably reduced due to the intrinsic property of high operator-dependence from US. In contrast, artificial intelligence (AI) excels at automatically recognizing complex patterns and providing quantitative assessment for imaging data, showing high potential to assist physicians in acquiring more accurate and reproducible results. In this article, we will provide a general understanding of AI, machine learning (ML) and deep learning (DL) technologies; We then review the rapidly growing applications of AI-especially DL technology in the field of US-based on the following anatomical regions: thyroid, breast, abdomen and pelvis, obstetrics heart and blood vessels, musculoskeletal system and other organs by covering image quality control, anatomy localization, object detection, lesion segmentation, and computer-aided diagnosis and prognosis evaluation; Finally, we offer our perspective on the challenges and opportunities for the clinical practice of biomedical AI systems in US.