Magnetic Resonance Imaging of Parotid Gland Tumors: Dynamic Contrast-Enhanced Sequence Evaluation

Machine learning-based radiomics for histological classification of parotid tumors using morphological MRI: a comparative study

OBJECTIVES

To evaluate the effectiveness of machine learning models based on morphological magnetic resonance imaging (MRI) radiomics in the classification of parotid tumors.

METHODS

In total, 298 patients with parotid tumors were randomly assigned to a training and test set at a ratio of 7:3. Radiomics features were extracted from the morphological MRI images and screened using the Select K Best and LASSO algorithm. Three-step machine learning models with XGBoost, SVM, and DT algorithms were developed to classify the parotid neoplasms into four subtypes. The ROC curve was used to measure the performance in each step. Diagnostic confusion matrices of these models were calculated for the test cohort and compared with those of the radiologists.

RESULTS

Six, twelve, and eight optimal features were selected in each step of the three-step process, respectively. XGBoost produced the highest area under the curve (AUC) for all three steps in the training cohort (0.857, 0.882, and 0.908, respectively), and for the first step in the test cohort (0.826), but produced slightly lower AUCs than SVM in the latter two steps in the test cohort (0.817 vs. 0.833, and 0.789 vs. 0.821, respectively). The total accuracies of XGBoost and SVM in the confusion matrices (70.8% and 59.6%) outperformed those of DT and the radiologist (46.1% and 49.2%).

CONCLUSION

This study demonstrated that machine learning models based on morphological MRI radiomics might be an assistive tool for parotid tumor classification, especially for preliminary screening in absence of more advanced scanning sequences, such as DWI.

KEY POINTS

• Machine learning algorithms combined with morphological MRI radiomics could be useful in the preliminary classification of parotid tumors. • XGBoost algorithm performed better than SVM and DT in subtype differentiation of parotid tumors, while DT seemed to have a poor validation performance. • Using morphological MRI only, the XGBoost and SVM algorithms outperformed radiologists in the four-type classification task for parotid tumors, thus making these models a useful assistant diagnostic tool in clinical practice.

MRI-based texture analysis to differentiate the most common parotid tumours

AIM

To evaluate magnetic resonance imaging (MRI) features and signal characteristics of parotid masses and investigate the added role of texture analysis (TA) in the differentiation of parotid tumours.

MATERIALS AND METHODS

Ninety-five patients (42 women, 53 men; mean age 51.67±14.15) were included in this study. The study group consisted of 40 pleomorphic adenoma, 45 Warthin's tumour, and 10 mucoepidermoid carcinomas. Two reviewers assessed the MRI sequences retrospectively. Fat-suppressed T2-weighted and contrast-enhanced T1-weighted axial images were used for TA. Receiver operating characteristic curve analyses were performed to evaluate the ability to make a diagnosis. Logistic regression analyses were conducted to explore the independent risk factors among the MRI features and to analyse the added value of TA to the qualitative analysis.

RESULTS

Significant differences were found in the tumour border (p<0.001), infiltration of the surrounding tissue (p=0.003), contrast-enhancement grading (p<0.001), perineural spread (p=0.013), and pathological lymph nodes (p<0.001) between the malignant and benign tumours. Kurtosis on contrast-enhanced T1-weighted images, and skewness and kurtosis on T2-weighted images were significantly different between the three groups (p=0.020, <0.001, 0.003; respectively). A kurtosis value on T2-weighted images <2.815 along with an ill-defined border had the highest specificity (98.8%) and positive predictive value (83.3%) in the differentiation of malignant tumours.

CONCLUSION

The addition of TA parameters to the MRI findings may contribute to distinguish benign from malignant parotid tumours.

Diagnostic accuracy of magnetic resonance imaging techniques for parotid tumors, a systematic review and meta-analysis

OBJECTIVE

To assess the added benefit of combining different MRI techniques for preoperative diagnosis of parotid tumors when compared to conventional MRI and advanced MRI techniques alone with meta-analysis.

METHODS

A comprehensive PubMed electronic database search was performed for original diagnostic studies up to July 2017. The methodologic quality of each study was evaluated by two independent reviewers who used the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool. Statistical analysis included pooling of sensitivity and specificity with 95% confidence intervals (CI). All analyses were conducted using STATA (version 12.0), RevMan (version 5.2), and Meta-Disc 1.4 software programs.

RESULTS

Pooled sensitivity and specificity of conventional MRI, diffusion weighted imaging (DWI), dynamic contrast enhanced (DCE) and the above combination were 76% (95%CI)/91% (95%CI)/80% (95%CI)/86% (95%CI) and 83% (95%CI)/56% (95%CI)/90% (95%CI)/90% (95%CI).

CONCLUSION

Conventional MRI combined with DWI and DCE showed higher diagnostic accuracy than conventional or advanced MRI alone, supporting their use in parotid tumors diagnosis.

Clinical utility of dynamic-enhanced MRI in salivary gland tumors: retrospective study and literature review

PURPOSE

To improve the diagnoses of the salivary gland tumors, a dynamic-enhanced MRI (dMRI) was investigated.

METHODS

We conducted a retrospective chart review of 93 cases of salivary gland tumors. The histological diagnoses were obtained from all patients using a surgical specimen and/or an open biopsy specimen. The dMRI as well as fine-needle aspiration cytology (FNAC) and intraoperative frozen section (IFS) were analyzed. This study focused on the time-intensity curve (TIC) after injection, peak time (Tpeak), washout ratio (WR) as well as the gradient of enhancement and washout profile.

RESULTS

The histological diagnoses included pleomorphic adenoma (PMA) in 53 cases, the Warthin tumors (WT) in 14 cases and malignant tumors (MT) in 26 cases. Incorrect diagnosis rate of FNAC and IFS were 5.2 and 8.3%, respectively. The TIC revealed differences among the three types of tumors. Tpeak as well as WR also revealed significant differences (p < 0.001). Tpeak were lower in order of WT, MT, PMA, respectively. WR of TICs at 30, 45 and 105 s after Tpeak were higher in order of WT, MT, PMA, respectively (p < 0.001). The gradient of increment and washout in the TIC curve was also an important parameter to distinguish the three types of tumors. In MT, the rapid enhancement pattern was found in high or intermediate histological grade tumors, whereas the slow enhancement pattern was exhibited in low grade tumors.

CONCLUSIONS

Our findings indicate that using Tpeak and WR, it is possible to distinguish between WT, PMA and MT. Additionally, a rapid enhancement pattern may be a potential marker for these tumors.