A Novel Sonographic Scoring Model in the Prediction of Major Salivary Gland Tumors

Machine Learning on Ultrasound Texture Analysis Data for Characterizing of Salivary Glandular Tumors: A Feasibility Study

BACKGROUND

Objective quantitative texture characteristics may be helpful in salivary glandular tumor differential diagnosis. This study uses machine learning (ML) to explore and validate the performance of ultrasound (US) texture features in diagnosing salivary glandular tumors.

MATERIAL AND METHODS

122 patients with salivary glandular tumors, including 71 benign and 51 malignant tumors, are enrolled. Representative brightness mode US pictures are selected for further Gray Level Co-occurrence Matrix (GLCM) texture analysis. We use a t-test to test the significance and use the receiver operating characteristic curve method to find the optimal cut-point for these significant features. After splitting 80% of the data into a training set and 20% data into a testing set, we use five machine learning models, k-nearest Neighbors (kNN), Naïve Bayes, Logistic regression, Artificial Neural Networks (ANNs) and supportive vector machine (SVM), to explore and validate the performance of US GLCM texture features in diagnosing salivary glandular tumors.

RESULTS

This study includes 49 female and 73 male patients, with a mean age of 53 years old, ranging from 21 to 93. We find that six GLCM texture features (contrast, inverse difference movement, entropy, dissimilarity, inverse difference and difference entropy) are significantly different between benign and malignant tumors (p < 0.05). In ML, the overall accuracy rates are 74.3% (95%CI: 59.8-88.8%), 94.3% (86.6-100%), 72% (54-89%), 84% (69.5-97.3%) and 73.5% (58.7-88.4%) for kNN, Naïve Bayes, Logistic regression, a one-node ANN and SVM, respectively.

CONCLUSIONS

US texture analysis with ML has potential as an objective and valuable tool to make a differential diagnosis between benign and malignant salivary gland tumors.

Ultrasonographic prediction model for benign and malignant salivary gland tumors: a preliminary study

OBJECTIVE

To establish an ultrasonographic (US) prediction model for benign and malignant salivary gland tumors.

STUDY DESIGN

We retrospectively analyzed the clinical data of 575 patients with salivary gland tumors. Patients were divided into benign (N = 420) and malignant (N = 155) tumor groups based on histopathologic results. The clinical and US features of the tumor groups were statistically compared. With histopathologic findings as the dependent variable and clinical and US features as independent variables, a multiple logistic regression model was established. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate its diagnostic efficacy.

RESULTS

Statistically significant differences between tumor groups were discovered for patient age, tumor site, and the US features of tumor size, shape, and margins; posterior echo pattern; microcalcification, abnormal lymph nodes, and tumor vascularity. Individual US features had limited diagnostic value. The AUC, sensitivity, specificity, and accuracy values of the logistic regression equation were 0.893, 84.3%, 80.0%, and 83.1%, respectively.

CONCLUSION

The diagnostic performance of the predictive model was significantly better than that of any single US factor. This suggests that establishment of multiple models based on US features can improve the accuracy of diagnosis of benign and malignant salivary gland tumors and can be applied clinically.

Comparisons among the Ultrasonography Prediction Model, Real-Time and Shear Wave Elastography in the Evaluation of Major Salivary Gland Tumors

We aimed to validate the diagnostic accuracy of a novel sonographic scoring model and compare it with other methods in the evaluation of malignant major salivary gland tumors. We enrolled 138 patients who received neck ultrasound (US) with fine needle aspiration (FNA) and further operations or core needle biopsies for major salivary gland tumors from June 2015 to October 2021. The sonographic scoring model was presented as 2.08 × (vague boundary) + 1.75 × (regional lymphadenopathy) + 1.18 × (irregular or speculated shape) + 1.45 × (absence of posterior acoustic enhancement) + 2.4 × (calcification). We compared the diagnostic accuracy of the sonographic scoring model with shear wave elastography (SWE), real-time elastography (RTE), and US-FNA cytology for differentiating between benign and malignant lesions. The sensitivity, specificity, and accuracy were 58%, 89%, and 85% for the sonographic scoring model, 74%, 62%, and 64% for the SD of SWE with optimal cutoff value of 31.5 kPa, 69%, 70%, and 70% for the 4-point scoring system of RTE, and 74%, 93%, and 91% for US-FNA cytology, respectively. The sonographic scoring model is feasible as assistance in the evaluation of major salivary gland tumors. US-FNA cytology remains the tool of choice in diagnosing malignant salivary gland tumors.

Calcification in Salivary Gland Cancer Mimicking Sialolithiasis-A Diagnostic Pitfall on Imaging: Report of Two Cases and Brief Review of the Literature

Objectives: Sialolithiasis is the most common cause of calcifications detected with ultrasound in patients with chronic inflammatory symptoms and swellings of the salivary glands. Other differential diagnoses of calcifications are extremely rare and mostly benign. Methods: Case report and literature review. Results: Two rare cases of malignant parotid gland tumors with calcifications in a localization typical for sialolithiasis, which were mistaken for salivary calculi based on image findings, are presented. Conclusions: This report intends to highlight the pitfalls in the imaging of parotid gland diseases. Even if malignant tumors of the parotid gland with calcifications are extremely rare, in ambiguous cases, differential diagnoses should be considered carefully. A high suspicion index of the need for further diagnostics in cases with calcifications is practical and could include missing periprandial symptoms, no obstruction signs in the proximal duct, and missing evidence of sialolithiasis in sialendoscopy.

Narrative review of multiparametric ultrasound in parotid gland evaluation

Disorders affecting parotid gland represent a heterogeneous group comprising congenital, inflammatory and neoplastic diseases which show a focal or diffuse pattern of appearance. The differentiation of neoplastic from non-neoplastic conditions of parotid glands is pivotal for the diagnostic imaging. Frequently there is evidence of overlapping between the clinical and the imaging appearance of the various pathologies. The parotid gland is also often object of study with the combination of different techniques [ultrasound-computed tomography-magnetic resonance imaging (US-CT-MRI), ex.]. Compared to other dominant methods of medical imaging, US has several advantages providing images in real-time at lower cost, and without harmful use of ionizing radiation and of contrast enhancement. B-mode US, and the microvascular pattern color Doppler are usually used as first step evaluation of parotid lesions. Elastography and contrast-enhanced US (CEUS) has opened further possible perspectives to improve the differentiation between benign and malignant parotid lesions. The characterization of the parotid tumors plays a crucial role for their treatment planning and for the prediction of possible surgical complications. We present, here an updated review of the most recurrent pathologies of parotid gland focusing on the diagnostic power of multiparametric US including CEUS and ultrasound elastography (USE); limitations, advantages and the main key-points will be presented.