A computer-aided diagnosing system in the evaluation of thyroid nodules-experience in a specialized thyroid center.

BACKGROUND: The evaluation of thyroid nodules with ultrasonography has created a large burden for radiologists. Artificial intelligence technology has been rapidly developed in recent years to reduce the cost of labor and improve the differentiation of thyroid malignancies. This study aimed to investigate the diagnostic performance of a novel computer-aided diagnosing system (CADs: S-detect) for the ultrasound (US) interpretation of thyroid nodule subtypes in a specialized thyroid center. METHODS: Our study prospectively included 180 thyroid nodules that underwent ultrasound interpretation. The CADs and radiologist assessed all nodules. The ultrasonographic features of different subtypes were analyzed, and the diagnostic performances of the CADs and radiologist were compared. RESULTS: There were seven subtypes of thyroid nodules, among which papillary thyroid cancer (PTC) accounted for 50.6% and follicular thyroid carcinoma (FTC) accounted for 2.2%. Among all thyroid nodules, the CADs presented a higher sensitivity and lower specificity than the radiologist (90.5% vs 81.1%; 41.2% vs 83.5%); the radiologist had a higher accuracy than the CADs (82.2% vs 67.2%) for diagnosing malignant thyroid nodules. The accuracy of the CADs was not as good as that of the radiologist in diagnosing PTCs (70.9% vs 82.1%). The CADs and radiologist presented accuracies of 43.8% and 60.9% in identifying FTCs, respectively. CONCLUSIONS: The ultrasound CADs presented a higher sensitivity for identifying malignant thyroid nodules than experienced radiologists. The CADs was not as good as experienced radiologists in a specialized thyroid center in identifying PTCs. Radiologists maintained a higher specificity than the CADs for FTC detection.

Size and Ultrasound Features Affecting Results of Ultrasound-Guided Fine-Needle Aspiration of Thyroid Nodules.

OBJECTIVES: Our goal was to assess the diagnostic efficacy of ultrasound (US)-guided fine-needle aspiration (FNA) of thyroid nodules according to size and US features. METHODS: A retrospective correlation was made with 1745 whole thyroidectomy and hemithyroidectomy specimens with preoperative US-guided FNA results. All cases were divided into 5 groups according to nodule size (≤5, 5.1-10, 10.1-15, 15.1-20, and >20 mm). For target nodules, static images and cine clips of conventional US and color Doppler were obtained. Ultrasound images were reviewed and evaluated by two radiologists with at least 5 years US working experience without knowing the results of pathology, and then agreement was achieved. RESULTS: The Bethesda category I rate was higher in nodules larger than 15 mm (P < .05). The diagnostic accuracy was best in nodules of 5 to 10 mm in diameter. The sensitivity, accuracy, PPV, and LR for negative US-guided FNA results were better in nodules with a size range of 5 to 15 mm. The specificity, negative predictive value (NPV), and LR for positive results and the Youden index rose with increasing nodule size. Seventeen false-positive and 60 false-negative results were found in this study. The false-negative rate rose with increasing nodule size. However, the false-positive rate was highest in the group containing the smallest nodules. Nodules with circumscribed margins and those that were nonsolid and nonhypoechoic and had no microcalcifications correlated with Bethesda I FNA results. Nodules with circumscribed margins and those that were nonsolid, heterogeneous, and nonhypoechoic and had increased vascularity correlated with false-negative FNA results. Borders correlated with Bethesda I false-negative and false-positive FNA results. CONCLUSIONS: Tiny nodules (≤5 mm) with obscure borders tended to yield false-positive FNA results. Large nodules (>20 mm) with several US features tended to yield false-negative FNA results.

Sonographic Features of Ductal Carcinoma In Situ of the Breast With Microinvasion: Correlation With Clinicopathologic Findings and Biomarkers.

OBJECTIVES: To compare the sonographic results, clinicopathologic characteristics, and biomarkers in pure ductal carcinoma in situ (DCIS) of the breast and DCIS with microinvasion. METHODS: A total of 218 patients with pathologically proven DCIS based on sonography in our hospital (2009-2013) were retrospectively enrolled. Clinicopathologic characteristics and biomarkers were examined. Grayscale sonographic results were investigated according to the American College of Radiology Breast Imaging Reporting and Data System lexicon, and color Doppler sonography was used to assess the vascularization distribution and degree. All variables were compared by univariate and multivariate logistic regression analyses. RESULTS: All patients were female, with a mean age of 55.3 years (range, 32-78 years). One hundred sixty patients with 160 lesions had pure DCIS, and 58 patients with 58 lesions had DCIS with microinvasion. Ductal carcinoma in situ with microinvasion was more likely to have sentinel lymph node metastases, larger tumors, a higher tumor grade, human epidermal growth factor receptor 2 positivity, and a high Ki-67 index (all P < .05). Univariate analysis showed that DCIS with microinvasion was more likely to be hypoechoic with microcalcifications, have a mixed vascularization distribution (equal peripheral and internal blood flow signals), and have a high degree of vascularization (at least 2 penetrating vessels; all P < .05). Multivariate analysis indicated that the presence of microcalcifications and a high degree of vascularization were significantly and independently associated with microinvasion (both P < .001). CONCLUSIONS: Our findings suggest that DCIS with microinvasion is more likely to have microcalcifications and a high degree of vascularization than pure DCIS. Patients with these sonographic features are more likely to have a high tumor grade, sentinel lymph node metastases, larger tumors, a high Ki-67 index, and human epidermal growth factor receptor 2 positivity.

Sonographic findings of localized Castleman disease of the abdomen and pelvis.

PURPOSE: The purpose of this study was to sonographically evaluate the diagnosis of localized Castleman disease in the abdomen and pelvis. METHODS: This was a retrospective analysis of 18 cases of Castleman disease localized in the abdomen and pelvis. The following features of the lesions were assessed on sonography (US): location, size, margin, echogenicity, echotexture, intralesional cystic necrosis, intralesional calcification, posterior acoustic enhancement, and blood supply. RESULTS: Of the 18 tumors, 16 were located in the abdomen and 2 were located in the pelvis close to iliac vessels. The most frequent appearance of localized Castleman disease in the abdomen and pelvis on US was of a single, well-defined, hypoechoic solid mass with no intralesional cystic necrosis. The internal echotexture was homogeneous in 4 cases and heterogeneous in 14 cases, with thin hyperechoic septa (n = 14) or calcifications (n = 3). Posterior acoustic enhancement was seen in 17 of the 18 cases (94%). Ninety-four percent of the lesions (17/18) had marked vascularity on color Doppler US. CONCLUSIONS: Localized Castleman disease in the abdomen and pelvis usually appears on US as a heterogeneously hypoechoic lesion containing thin septa, and more commonly than not, demonstrates posterior acoustic enhancement and marked vascularity.

Predictive nomogram using multimodal ultrasonographic features for axillary lymph node metastasis in early­stage invasive breast cancer.

Axillary lymph node (ALN) status is a key prognostic factor in patients with early-stage invasive breast cancer (IBC). The present study aimed to develop and validate a nomogram based on multimodal ultrasonographic (MMUS) features for early prediction of axillary lymph node metastasis (ALNM). A total of 342 patients with early-stage IBC (240 in the training cohort and 102 in the validation cohort) who underwent preoperative conventional ultrasound (US), strain elastography, shear wave elastography and contrast-enhanced US examination were included between August 2021 and March 2022. Pathological ALN status was used as the reference standard. The clinicopathological factors and MMUS features were analyzed with uni- and multivariate logistic regression to construct a clinicopathological and conventional US model and a MMUS-based nomogram. The MMUS nomogram was validated with respect to discrimination, calibration, reclassification and clinical usefulness. US features of tumor size, echogenicity, stiff rim sign, perfusion defect, radial vessel and US Breast Imaging Reporting and Data System category 5 were independent risk predictors for ALNM. MMUS nomogram based on these factors demonstrated an improved calibration and favorable performance [area under the receiver operator characteristic curve (AUC), 0.927 and 0.922 in the training and validation cohorts, respectively] compared with the clinicopathological model (AUC, 0.681 and 0.670, respectively), US-depicted ALN status (AUC, 0.710 and 0.716, respectively) and the conventional US model (AUC, 0.867 and 0.894, respectively). MMUS nomogram improved the reclassification ability of the conventional US model for ALNM prediction (net reclassification improvement, 0.296 and 0.288 in the training and validation cohorts, respectively; both P<0.001). Taken together, the findings of the present study suggested that the MMUS nomogram may be a promising, non-invasive and reliable approach for predicting ALNM.

Machine Learning-Based Breast Tumor Ultrasound Radiomics for Pre-operative Prediction of Axillary Sentinel Lymph Node Metastasis Burden in Early-Stage Invasive Breast Cancer.

OBJECTIVE: The aim of the work described here was to assess the application of ultrasound (US) radiomics with machine learning (ML) classifiers to the prediction of axillary sentinel lymph node metastasis (SLNM) burden in early-stage invasive breast cancer (IBC). METHODS: In this study, 278 early-stage IBC patients with at least one SLNM (195 in the training set and 83 in the test set) were studied at our institution. Pathologic SLNM burden was used as the reference standard. The US radiomics features of breast tumors were extracted by using 3D-Slicer and PyRadiomics software. Four ML classifiers-linear discriminant analysis (LDA), support vector machine (SVM), random forest (RF) and decision tree (DT)-were used to construct radiomics models for the prediction of SLNM burden. The combined clinicopathologic-radiomics models were also assessed with respect to sensitivity, specificity, accuracy and areas under the curve (AUCs). RESULTS: Among the US radiomics models, the SVM classifier achieved better predictive performance with an AUC of 0.920 compared with RF (AUC = 0.874), LDA (AUC = 0.835) and DT (AUC = 0.800) in the test set. The clinicopathologic model had low efficacy, with AUCs of 0.678 and 0.710 in the training and test sets, respectively. The combined clinicopathologic (C) factors and SVM classifier (C + SVM) model improved the predictive ability with an AUC of 0.934, sensitivity of 86.7%, specificity of 89.9% and accuracy of 91.0% in the test set. CONCLUSION: ML-based US radiomics analysis, as a novel and promising predictive tool, is conducive to a precise clinical treatment strategy.

Association of sonographic features and molecular subtypes in predicting breast cancer disease outcomes.

BACKGROUND: Features in preoperative ultrasound could predict the prognosis of triple-negative breast cancer (TNBC), while its prognostic value in other molecular subtypes of breast cancer (BC) was unknown. The study aimed to assess the prognostic value of preoperative sonographic features, including orientations, on long-term outcomes in BC and its association with different molecular subtypes. METHODS: Women diagnosed with invasive BC > 5 mm who underwent surgery were retrospectively reviewed. Clinical, pathological, and sonographic profiles were collected and recurrence-free survival (RFS) and breast cancer-specific survival (BCSS) were reported. Interactions between clinicopathological features and tumor orientations in predicting RFS and BCSS were analyzed. Competing risk model was performed to estimate prognostic values of sonographic features for RFS and BCSS. RESULTS: A total of 2812 patients were included. With a median follow-up of 60.0 months, 268 (9.5%) patients suffered from recurrences and 104 (3.7%) died of BC. The prognostic values of vertical orientation in predicting RFS (P = .001) and BCSS (P = .001) were strongly associated with molecular subtypes. Non-TNBC tumors with vertical orientation had less recurrence events compared with parallel orientation (6.3% vs 8.7%, P = .035), whereas failed to predict disease outcomes in multivariate analysis (P > .05). Oppositely, in TNBC, vertical orientation was associated with worse RFS (HR = 3.50; 95% confidence interval [CI] 1.69-7.24; P < .001) and BCSS (HR = 6.36; 95% CI 2.86-14.14; P < .001) in multivariate analysis with a 5-year RFS and BCSS of 73.4% and 74.6%. Meanwhile, vertical orientation was related with smaller tumor size (P < .001), human epidermal growth factor receptor 2 nonamplification (P < .001), and lower Ki-67 expression (P = .001) among non-TNBC population, whereas TNBC tumors with vertical orientation had a higher burden of axillary lymph node metastases (2.8 ± 1.0 vs 1.4 ± 0.2, P = .001). CONCLUSIONS: Prognostic values of sonographic orientation in predicting BC disease outcomes were associated with molecular subtypes. Vertical orientation in preoperative sonogram may serve as a prognostic biomarker for TNBC patients.

Clinical characteristics and ultrasonographic features for predicting central lymph node metastasis in clinically node-negative papillary thyroid carcinoma without capsule invasion.

BACKGROUND: To evaluate the association of preoperative clinical and sonographic features with central lymph node metastasis (CLNM) in patients with clinically node-negative (cN0) papillary thyroid carcinoma (PTC) without capsule invasion. METHODS: Clinical and sonographic features of 635 cN0 PTC nodules without capsule invasion were retrospectively reviewed. CLNM was confirmed by pathology. Univariate and multivariate analyses were performed to analyze the predicting factors associated with CLNM in cN0 PTC without capsule invasion. RESULTS: In the 635 cN0 PTC nodules without capsule invasion, age ≤36 years, male, tumor size >8 mm, the distance between the tumor and the capsule ≤1.1 mm and calcification were independently associated with CLNM (P < .05). CONCLUSIONS: CLNM was associated with age, sex, tumor size, the distance between the tumor and the capsule, and calcification in cN0 PTC without capsule invasion. Preoperative assessment of risk factors could help to select PTC patients who benefit from surgery.

Comparison of fine-needle aspiration and fine-needle capillary sampling of thyroid nodules: a prospective study with emphasis on the influence of nodule size.

BACKGROUND: The objective of this study was to compare the sampling efficiency of ultrasound-guided fine-needle aspiration (FNA) and fine-needle capillary (FNC) sampling in thyroid nodules, in which the authors specifically analyzed the influence of nodule size. METHODS: This study included 280 thyroid nodules in 275 consecutive patients. The nodules were divided into 4 size subgroups: ≤5.0 mm, from 5.1 to 10.0 mm, from 10.1 to 20.0 mm, and >20.0 mm. Each nodule was sampled by both FNA and FNC. The final cytopathologic findings were reported. The smears were scored and then categorized as diagnostically inadequate, adequate, or superior on the basis of 4 parameters, which included background clot or blood, the number of obtained cells, preserved tissue architecture, and cellular degeneration. RESULTS: The κ scores for agreement of the cytopathologic results between FNA and FNC sampling in the 4 size subgroups were 0.377, 0.455, 0.751, and 0.352 for nodules that measured ≤5.0 mm, from 5.1 to 10.0 mm, from 10.1 to 20.0 mm, and >20.0 mm, respectively. The proportion of nondiagnostic of FNAs was significantly lower than the proportion of nondiagnostic FNC samples in nodules that measured >20.0 mm (P = .037). Scores for the 4 diagnostic parameters were significantly greater in FNAs than in FNC samples in nodules that measured from 5.1 to 10.0 mm and >20.0 mm (all P < .05); however, similar results were not observed in the nodules that measured ≤5.0 mm or from 10.1 to 20.0 mm (all P > .05). Also, FNA yielded significantly more diagnostically superior specimens than FNC sampling in nodules that measured from 5.1 to 10.0 mm and >20.0 mm (P < .05 for both). CONCLUSIONS: The current findings indicated that FNA may be more suitable than FNC for sampling nodules that measure from 5.1 to 10.0 mm and >20.0 mm; whereas, for nodules that measure ≤5.0 mm and from 10.1 to 20.0 mm, the 2 techniques could yield specimens with similar quality.