A mathematical model using computed tomography for the diagnosis of metastatic central compartment lymph nodes in papillary thyroid carcinoma

Dual-source dual-energy CT and deep learning for equivocal lymph nodes on CT images for thyroid cancer

OBJECTIVES

This study investigated the diagnostic performance of dual-energy computed tomography (CT) and deep learning for the preoperative classification of equivocal lymph nodes (LNs) on CT images in thyroid cancer patients.

METHODS

In this prospective study, from October 2020 to March 2021, 375 patients with thyroid disease underwent thin-section dual-energy thyroid CT at a small field of view (FOV) and thyroid surgery. The data of 183 patients with 281 LNs were analyzed. The targeted LNs were negative or equivocal on small FOV CT images. Six deep-learning models were used to classify the LNs on conventional CT images. The performance of all models was compared with pathology reports.

RESULTS

Of the 281 LNs, 65.5% had a short diameter of less than 4 mm. Multiple quantitative dual-energy CT parameters significantly differed between benign and malignant LNs. Multivariable logistic regression analyses showed that the best combination of parameters had an area under the curve (AUC) of 0.857, with excellent consistency and discrimination, and its diagnostic accuracy and sensitivity were 74.4% and 84.2%, respectively (p < 0.001). The visual geometry group 16 (VGG16) based model achieved the best accuracy (86%) and sensitivity (88%) in differentiating between benign and malignant LNs, with an AUC of 0.89.

CONCLUSIONS

The VGG16 model based on small FOV CT images showed better diagnostic accuracy and sensitivity than the spectral parameter model. Our study presents a noninvasive and convenient imaging biomarker to predict malignant LNs without suspicious CT features in thyroid cancer patients.

CLINICAL RELEVANCE STATEMENT

Our study presents a deep-learning-based model to predict malignant lymph nodes in thyroid cancer without suspicious features on conventional CT images, which shows better diagnostic accuracy and sensitivity than the regression model based on spectral parameters.

KEY POINTS

Many cervical lymph nodes (LNs) do not express suspicious features on conventional computed tomography (CT). Dual-energy CT parameters can distinguish between benign and malignant LNs. Visual geometry group 16 model shows superior diagnostic accuracy and sensitivity for malignant LNs.

Integrative lymph node-mimicking models created with biomaterials and computational tools to study the immune system

The lymph node (LN) is a vital organ of the lymphatic and immune system that enables timely detection, response, and clearance of harmful substances from the body. Each LN comprises of distinct substructures, which host a plethora of immune cell types working in tandem to coordinate complex innate and adaptive immune responses. An improved understanding of LN biology could facilitate treatment in LN-associated pathologies and immunotherapeutic interventions, yet at present, animal models, which often have poor physiological relevance, are the most popular experimental platforms. Emerging biomaterial engineering offers powerful alternatives, with the potential to circumvent limitations of animal models, for in-depth characterization and engineering of the lymphatic and adaptive immune system. In addition, mathematical and computational approaches, particularly in the current age of big data research, are reliable tools to verify and complement biomaterial works. In this review, we first discuss the importance of lymph node in immunity protection followed by recent advances using biomaterials to create in vitro/vivo LN-mimicking models to recreate the lymphoid tissue microstructure and microenvironment, as well as to describe the related immuno-functionality for biological investigation. We also explore the great potential of mathematical and computational models to serve as in silico supports. Furthermore, we suggest how both in vitro/vivo and in silico approaches can be integrated to strengthen basic patho-biological research, translational drug screening and clinical personalized therapies. We hope that this review will promote synergistic collaborations to accelerate progress of LN-mimicking systems to enhance understanding of immuno-complexity.

Dual-source dual-energy thin-section CT combined with small field of view technique for small lymph node in thyroid cancer: a retrospective diagnostic study

Background

To evaluate the diagnostic performance of quantitative spectral parameters derived from dual-source dual-energy CT at small field of view (FOV) for small lymph node metastasis in thyroid cancer.

Methods

This was a retrospective diagnostic study. From 2016 to 2019, 280 patients with thyroid disease underwent thin-section dual-source dual-energy thyroid CT and thyroid surgery. The data of patients with lymph nodes having a short diameter of 2-6 mm was analyzed. The quantitative dual-energy CT parameters of targeted lymph nodes were measured, and all parameters between metastatic and non-metastatic lymph nodes were compared. These parameters were then fitted to univariable and multivariable binary logistic regression models. The diagnostic role of spectral parameters was analyzed by receiver operating characteristic (ROC) curves and compared with the McNemar test. Small FOV CT images and a mathematical model were used to judge the status of lymph nodes respectively, and then compared with the golden criterion-pathological diagnosis. The cut-off value of the model was 0.4419, with a sensitivity of 90.2% and a specificity of 92.7%.

Results

Of the 216 lymph nodes investigated in this study, 52.3% and 23.6% had a short diameter of 2-3 and 4 mm, respectively. Multiple quantitative CT parameters were significantly different between benign and malignant lymph nodes, and binary regression analysis was performed. The mathematical model was: p=ey/(1+ ey), y=-23.119+0.033× precontrast electron cloud density +0.076× arterial phase normalized iodine concentration +2.156× arterial phase normalized effective atomic number -0.540× venous phase slope of the spectral Hounsfield unit curve +1.676× venous phase iodine concentration. This parameter model had an AUC of 92%, with good discrimination and consistency, and the diagnostic accuracy was 90.3%. The diagnostic accuracy of the CT image model was 43.1%, and for lymph nodes with a short-diameter of 2-3 mm, the diagnostic accuracy was 22.1%.

Conclusions

The parameter model showed higher diagnostic accuracy than the CT image model for diagnosing small lymph node metastasis in thyroid cancer, and quantitative dual-energy CT parameters were very useful for small lymph nodes that were difficult to be diagnosed only on conventional CT images.

Trial registration

This study is retrospectively registered, and we have registered a prospective study (Registration number: ChiCTR2000035195; http://www.chictr.org.cn).

Development and validation of a nomogram for preoperative prediction of level VII nodal spread in papillary thyroid cancer: Radiologic-pathologic correlation

PURPOSE

To develop and validate a diagnostic nomogram for preoperative prediction of the level VII nodal spread in papillary thyroid cancer (PTC) by incorporating CT features.

METHODS

A dataset of 7896 patients experiencing thyroidectomy for thyroid cancer was collected retrospectively from two hospitals, and 300 patients were finally included in this study. The CT features of metastatic LN were extracted with a one by one match of radiologic-pathologic correlation. Multivariable binary logistic regression analysis was used to develop predicting model, and then a nomogram was developed utilizing a primary cohort of 152 patients from hospital #1. The nomogram was validated in external cohort of 62 patients from hospital #2 and an independent cohort of 86 patients from hospital #1. The performance of the nomogram was evaluated with respect to its calibration, discrimination.

RESULTS

531 LNs from 300 patients were analyzed. 42.6% LNs were > 5 mm in short diameter. A total of 7 selected CT features were significantly associated with LN status (P < 0.05), including nodular enhancement, cystic change, calcification and so on. These features were contained in the prediction nomogram. The model showed good discrimination and good calibration, with a C-index of 0.938 (95% CI, 0.913 to 0.963) and 0. 795 (95% CI, 0. 726 to 0.864) for the primary cohort and the validation cohort, respectively. Decision curve analysis demonstrated that the nomogram was clinically applicable.

CONCLUSIONS

This nomogram incorporating pathologically relevant CT features has demonstrated a high diagnostic value for predicting level VII nodal spread in PTC. Our work may help thyroid surgeon to decide whether upper mediastinal lymphadenectomy should be performed, which is associated with thoracotomy or other surgery.

Risk model and risk stratification to preoperatively predict central lymph node metastasis in papillary thyroid carcinoma

Background

The central lymph node is the most common involvement for papillary thyroid carcinoma (PTC), which is correlated to recurrence and survival. But it is difficult to accurately evaluate lymph node prior to an operation. This retrospective study was designed to develop a risk model and risk stratification to preoperatively predict central lymph node metastasis (CLNM) in PTC and validate this model.

Methods

A series of 1,714 initial treatment PTC patients were enrolled. Among these patients, 1,001 patients were used to develop a predictive model and establish a stratification scoring system. This was validated through the remaining 713 patients.

Results

The multivariate analysis revealed that CLNM and lateral lymph node metastasis (LLNM) in ultrasound (US), tumor size, gender, capsule invasion in US, microcalcification and age were significant independent predictors for CLNM. The area under the curve (AUC) of the model was 0.778. Furthermore, the cutoff value to predict CLNM was 8 points, and the sensitivity and specificity were 77% and 65%, respectively. In the scoring system for CLNM, a score of ≤8, 8-18 and >18 were defined as low, intermediate and high risk, respectively. The risk of CLNM was approximately 30%, 60% and 80%, corresponding to the stratification. When validated, the model predicted the risk of CLNM with an AUC of 0.811, a sensitivity and specificity of 83% and 63%, respectively.

Conclusions

This study presented a predictive model to preoperatively assess the risk of CLNM in PTC. The predictive model performed well, but needed to be prospectively validated in external center.

Clinical Validation of a Predictive Model for the Presence of Cervical Lymph Node Metastasis in Papillary Thyroid Cancer

BACKGROUND AND PURPOSE

Ultrasound is a standard technique to detect lymph node metastasis in papillary thyroid cancer. Cystic changes and microcalcifications are the most specific features of metastasis, but with low sensitivity. This prospective study compared the diagnostic accuracy of a predictive model for sonographic evaluation of lymph nodes relative to the radiologist's standard assessment in detecting papillary thyroid cancer metastasis in patients after thyroidectomy.

MATERIALS AND METHODS

Cervical lymph node sonographic images were reported by a radiologist (R method) per standard practice. The same images were independently evaluated by another radiologist using a sonographic predictive model (M method). A test was considered positive for metastasis if the R or M method suggested lymph node biopsy. The result of lymph node biopsy or surgical pathology was used as the reference standard. We estimated relative true-positive fraction and relative false-positive fraction using log-linear models for correlated binary data for the M method compared with the R method.

RESULTS

A total of 237 lymph nodes in 103 patients were evaluated. Our analysis of relative true-positive fraction and relative false-positive fraction included 54 nodes with pathologic results in which at least 1 method (R or M) was positive. The M method had a higher relative true-positive fraction of 1.46 (95% CI, 1.12-1.91; P = .006) and a lower relative false-positive fraction of 0.58 (95% CI, 0.36-0.92; P = .02) compared with the R method.

CONCLUSIONS

The sonographic predictive model outperformed the standard assessment to detect lymph node metastasis in patients with papillary thyroid cancer and may reduce unnecessary biopsies.