The role of three-dimensional pseudo-continuous arterial spin labelling in grading and differentiating histological subgroups of meningiomas

Meningioma grading via diagnostic imaging: A systematic review and meta-analysis

PURPOSE

Meningioma is the most common intracranial tumor, graded on pathology using WHO criteria to predict tumor course and treatment. However, pathological grading via biopsy may not be possible in cases with poor surgical access due to tumor location. Therefore, our systematic review aims to evaluate whether diagnostic imaging features can differentiate high grade (HG) from low grade (LG) meningiomas as an alternative to pathological grading.

METHODS

Three databases were searched for primary studies that either use routine magnetic resonance imaging (MRI) or computed tomography (CT) to assess pathologically WHO-graded meningiomas. Two investigators independently screened and extracted data from included studies.

RESULTS

24 studies met our inclusion criteria with 12 significant (p < 0.05) CT and MRI features identified for differentiating HG from LG meningiomas. Cystic changes in the tumor had the highest specificity (93.4%) and irregular tumor-brain interface had the highest positive predictive value (65.0%). Mass effect had the highest sensitivity (81.0%) and negative predictive value (90.7%) of all imaging features. Imaging feature with the highest accuracy for identifying HG disease was irregular tumor-brain interface (79.7%). Irregular tumor-brain interface and heterogenous tumor enhancement had the highest AUC values of 0.788 and 0.703, respectively.

CONCLUSION

Our systematic review highlight imaging features that can help differentiate HG from LG meningiomas.

A radiomics-based study for differentiating parasellar cavernous hemangiomas from meningiomas

To investigate the value of the radiomic models for differentiating parasellar cavernous hemangiomas from meningiomas and to compare the classification performance with different MR sequences and classifiers. A total of 96 patients with parasellar tumors (40 cavernous hemangiomas and 56 meningiomas) were enrolled in this retrospective multiple-center study. Univariate and multivariate analyses were performed to identify the clinical factors and semantic features of MRI scans. Radiomics features were extracted from five MRI sequences using radiomics software. Three feature selection methods and six classifiers were evaluated in the training cohort to construct favorable radiomic machine-learning classifiers. The performance of different classifiers was evaluated using the AUC and compared to neuroradiologists. The detection rates of T1WI, T2WI, and CE-T1WI for parasellar cavernous hemangiomas and meningiomas were approximately 100%. In contrast, the ADC maps had the detection rate of 18/22 and 19/25, respectively, (AUC, 0.881) with 2.25 cm as the critical value diameter. Radiomics models with the SVM and KNN classifiers based on T2WI and ADC maps had favorable predictive performances (AUC > 0.90 and F-score value > 0.80). These models outperformed MRI model (AUC 0.805) and neuroradiologists (AUC, 0.756 and 0.545, respectively). Radiomic models based on T2WI and ADC and combined with SVM and KNN classifiers have the potential to be a viable method for differentiating parasellar hemangiomas from meningiomas. T2WI is more universally applicable than ADC values due to its higher detection rate for parasellar tumors.

Biomarkers for differentiating grade II meningiomas from grade I: a systematic review

INTRODUCTION

There are a number of prognostic markers (methylation, CDKN2A/B) described to be useful for the stratification of meningiomas. However, there are currently no clinically validated biomarkers for the preoperative prediction of meningioma grade, which is determined by the histological analysis of tissue obtained from surgery. Accurate preoperative biomarkers would inform the pre-surgical assessment of these tumours, their grade and prognosis and refine the decision-making process for treatment. This review is focused on the more controversial grade II tumours, where debate still surrounds the need for adjuvant therapy, repeat surgery and frequency of follow up.

METHODS

We evaluated current literature for potential grade II meningioma clinical biomarkers, focusing on radiological, biochemical (blood assays) and immunohistochemical markers for diagnosis and prognosis, and how they can be used to differentiate them from grade I meningiomas using the post-2016 WHO classification. To do this, we conducted a PUBMED, SCOPUS, OVID SP, SciELO, and INFORMA search using the keywords; 'biomarker', 'diagnosis', 'atypical', 'meningioma', 'prognosis', 'grade I', 'grade 1', 'grade II' and 'grade 2'.

RESULTS

We identified 1779 papers, 20 of which were eligible for systematic review according to the defined inclusion and exclusion criteria. From the review, we identified radiological characteristics (irregular tumour shape, tumour growth rate faster than 3cm³/year, high peri-tumoural blood flow), blood markers (low serum TIMP1/2, high serum HER2, high plasma Fibulin-2) and histological markers (low H3K27me3, low SMARCE1, low AKAP12, high ARIDB4) that may aid in differentiating grade II from grade I meningiomas.

CONCLUSION

Being able to predict meningioma grade at presentation using the radiological and blood markers described may influence management as the likely grade II tumours will be followed up or treated more aggressively, while the histological markers may prognosticate progression or post-treatment recurrence. This to an extent offers a more personalised treatment approach for patients.

Superselective pseudocontinuous arterial spin labeling in patients with meningioma: utility in prediction of feeding arteries and preoperative embolization feasibility

OBJECTIVE

Superselective pseudocontinuous arterial spin labeling (ss-pCASL) is an MRI technique in which individual vessels are labeled to trace their perfusion territories. In this study, the authors assessed its merit in defining feeding vessels and gauging preoperative embolization feasibility for patients with meningioma, using digital subtraction angiography (DSA) as the reference method.

METHODS

Thirty-one consecutive patients with meningiomas were prospectively recruited, each undergoing DSA (and embolization, if feasible) before resection. All ss-pCASL imaging studies were performed 1 day prior to DSA. Two neuroradiologists independently reviewed ss-pCASL images, rating the contribution of each labeled vessel to tumor blood supply as none, minor, or major. Two neuroradiologists also gauged the feasibility of embolization in each patient, based on ss-pCASL images. Interobserver and intermodality agreement were determined using Cohen's kappa statistic. The diagnostic performance of ss-pCASL was assessed in terms of discerning tumor blood supply and the potential for embolization.

RESULTS

Interobserver agreement in the rating of blood supply by ss-pCASL was very good (κ = 0.817, 95% CI 0.771-0.863), and intermodality agreement (consensus ss-pCASL readings vs DSA findings) was good (κ = 0.688, 95% CI 0.632-0.744). In delineating tumor blood supply, ss-pCASL showed high sensitivity (87.1%) and specificity (87.2%). The positive and negative predictive values for embolization feasibility were 85.2% and 100%, respectively.

CONCLUSIONS

In patients with meningiomas, feeding vessels are reliably predicted by ss-pCASL. This noninvasive approach, involving no iodinated contrast or radiation exposure, is particularly beneficial if there are no prospects of embolization.

The Current State of Radiomics for Meningiomas: Promises and Challenges

Meningiomas are the most common primary tumors of the central nervous system. Given the fact that the majority of meningiomas are benign, the preoperative risk stratification and treatment strategy decision-making highly rely on the conventional subjective radiologic evaluation. However, this traditional diagnostic and treatment modality may not be effective in patients with aggressive-growing tumors or symptomatic patients with potential risk of recurrence after surgical resection or radiotherapy, as this passive “wait and see” strategy could miss the optimal opportunity of intervention. Radiomics, a new rising discipline, translates high-dimensional image information into abundant mathematical data by multiple computational algorithms. It provides an objective and quantitative approach to interpret the imaging data, rather than the subjective and qualitative interpretation from relatively limited human visual observation. In fact, the enormous amount of information generated by radiomics analyses provides radiological to histopathological tumor information, which are visually imperceptible, and offers technological basis to its applications amid diagnosis, treatment, and prognosis. Here, we review the latest advancements of radiomics and its applications in the prediction of the pathological grade, pathological subtype, recurrence possibility, and differential diagnosis of meningiomas, and the potential and challenges in general clinical applications. In this review, we highlight the generalization of shared radiomic features among different studies and compare different performances of popular algorithms. At last, we discuss several possible aspects of challenges and future directions in the development of radiomic applications in meningiomas.

Machine Learning in Meningioma MRI: Past to Present. A Narrative Review

Meningioma is one of the most frequent primary central nervous system tumors. While magnetic resonance imaging (MRI), is the standard radiologic technique for provisional diagnosis and surveillance of meningioma, it nevertheless lacks the prima facie capacity in determining meningioma biological aggressiveness, growth, and recurrence potential. An increasing body of evidence highlights the potential of machine learning and radiomics in improving the consistency and productivity and in providing novel diagnostic, treatment, and prognostic modalities in neuroncology imaging. The aim of the present article is to review the evolution and progress of approaches utilizing machine learning in meningioma MRI-based sementation, diagnosis, grading, and prognosis. We provide a historical perspective on original research on meningioma spanning over two decades and highlight recent studies indicating the feasibility of pertinent approaches, including deep learning in addressing several clinically challenging aspects. We indicate the limitations of previous research designs and resources and propose future directions by highlighting areas of research that remain largely unexplored. LEVEL OF EVIDENCE: 5 TECHNICAL EFFICACY STAGE: 2.

The Diagnostic Value of Radiomics-Based Machine Learning in Predicting the Grade of Meningiomas Using Conventional Magnetic Resonance Imaging: A Preliminary Study

Objective: The purpose of the current study is to investigate whether texture analysis-based machine learning algorithms could help devise a non-invasive imaging biomarker for accurate classification of meningiomas using machine learning algorithms.

Method: The study cohort was established from the hospital database by reviewing the medical records. Patients were selected if they underwent meningioma resection in the neurosurgery department between January 2015 and December 2018. A total number of 40 texture parameters were extracted from pretreatment postcontrast T1-weighted (T1C) images based on six matrixes. Three feature selection methods were adopted, namely, distance correlation, least absolute shrinkage and selection operator (LASSO), and gradient boosting decision tree (GBDT). Multiclass classification methods of linear discriminant analysis (LDA) and support vector machine (SVM) algorithms were employed to establish the classification models. The diagnostic performances of models were evaluated with confusion matrix based on which the areas under the curve, accuracy, and Kappa value of models were calculated.

Result: Confusion matrix showed that the LDA-based models represented better diagnostic performances than SVM-based models. The highest accuracy among LDA-based models was 75.6%, shown in the combination of Lasso + LDA. The optimal models for SVM-based models was Lasso+SVM, with accuracy of 59.0% in the testing group. One of the SVM-based models, GBDT+SVM, was overfitting, suggesting that this model was not suitable for application.

Conclusion: Machine learning algorithms with texture features extracted from T1C images could potentially serve as the assistant imaging biomarkers for presurgically grading meningiomas.

Adhesive and non-adhesive internal hernia: clinical relevance and multi-detector CT images

Internal hernia (IH)-related surgical acute abdomen is not well understood because of the rarity of cases and underdiagnosis. This study was performed to further understand the clinicopathological features and multi-detector computed tomography (MDCT) findings of IH in cases confirmed by surgery. In all, 51 patients with a definite diagnosis of IH confirmed during surgical exploration from Feb. 2012 to Feb. 2018 in our hospital were included in this research. Medical records, including MDCT images and intra-operative findings, were collected retrospectively. In all, 39 and 12 cases were categorized as adhesive IH (76.5%) and non-adhesive IH (23.5%), respectively. Among the patients with adhesive IH, 73% had a history of abdominal or pelvic surgery. Additionally, the mesentery was the most common component of adhesive bands (64.1%). Congenital peritoneal abnormalities and gastrointestinal reconstruction were the main causes of non-adhesive IH.As a specific sign, the fat notch sign was much more common in adhesive IH than in non-adhesive IH (P = 0.023). Bowel wall thickening (P = 0.041), abnormal bowel wall enhancement (P = 0.006) and twisted bowels with the vessel swirl sign (P = 0.004) were indicators of bowel necrosis. Among all of the cases of IH, 34 (66.7%) were complicated by bowel necrosis, and 1 patient died. In conclusion, non-adhesive IH has different clinicopathological features and MDCT findings from those of adhesive IH. MDCT is a useful tool with high sensitivity for confirming IH and may help to guide the early treatment of IH.

Can amide proton transfer-weighted imaging differentiate tumor grade and predict Ki-67 proliferation status of meningioma?

OBJECTIVES

To determine the utility of the amide proton transfer-weighted MR imaging in differentiating the WHO grade and predict proliferative activity of meningioma.

METHODS

Fifty-three patients with WHO grade I meningiomas and 26 patients with WHO grade II meningiomas underwent conventional and APT-weighted sequences on a 3.0 Tesla MR before clinical intervention. The APT-weighted (APTw) parameters in the solid tumor region were obtained and compared between two grades using the t test; the receiver operating characteristic (ROC) curve was used to assess the best parameter for predicting the grade of meningiomas. Pearson's correlation coefficient was calculated between the APTw_max and Ki-67 labeling index in meningiomas.

RESULTS

The APTw_max and APTw_mean values were not significantly different between WHO grade I and grade II meningiomas (p = 0.103 and p = 0.318). The APTw_min value was higher and the APTw_max-min value was lower in WHO grade II meningiomas than in WHO grade I tumors (p = 0.027 and p = 0.019). But the APTw_min was higher and the APTw_max-min was lower in microcystic meningiomas than in WHO grade II meningiomas (p = 0.001 and p = 0.006). The APTw_min combined with APTw_max-min showed the best diagnostic performance in predicting the grade of meningiomas with an AUC of 0.772. The APTw_max value was positively correlated with Ki-67 labeling index (r = 0.817, p < 0.001) in meningiomas; the regression equation for the Ki-67 labeling index (%) (Y) and APTw_max (%) (X) was Y = 4.9 × X - 12.4 (R² = 0.667, p < 0.001).

CONCLUSION

As a noninvasive imaging method, the ability of APTw-MR imaging in differentiating the grade of meningiomas is limited, but the technology can be used to predict the proliferative activity of meningioma.

KEY POINTS

• The APTw _min value was higher and the APTw _max-min value was lower in WHO grade II meningioma than in grade I tumors. • The APTw _min value was higher and the APTw _max-min value was lower in microcystic meningiomas than in WHO grade II meningiomas. • The APTw _max value was positively correlated with meningioma proliferation index.

The diagnostic value of texture analysis in predicting WHO grades of meningiomas based on ADC maps: an attempt using decision tree and decision forest

OBJECTIVES

The preoperative prediction of the WHO grade of a meningioma is important for further treatment plans. This study aimed to assess whether texture analysis (TA) based on apparent diffusion coefficient (ADC) maps could non-invasively classify meningiomas accurately using tree classifiers.

METHODS

A pathology database was reviewed to identify meningioma patients who underwent tumour resection in our hospital with preoperative routine MRI scanning and diffusion-weighted imaging (DWI) between January 2011 and August 2017. A total of 152 meningioma patients with 421 preoperative ADC maps were included. Four categories of features, namely, clinical features, morphological features, average ADC values and texture features, were extracted. Three machine learning classifiers, namely, classic decision tree, conditional inference tree and decision forest, were built on these features from the training dataset. Then the performance of each classifier was evaluated and compared with the diagnosis made by two neuro-radiologists.

RESULTS

The ADC value alone was unable to distinguish three WHO grades of meningiomas. The machine learning classifiers based on clinical, morphological features and ADC value could achieve equivalent diagnostic performance (accuracy = 62.96%) compared to two experienced neuro-radiologists (accuracy = 61.11% and 62.04%). Upon analysis, the decision forest that was built with 23 selected texture features and the ADC value from the training dataset achieved the best diagnostic performance in the testing dataset (kappa = 0.64, accuracy = 79.51%).

CONCLUSIONS

Decision forest with the ADC value and ADC map-based texture features is a promising multiclass classifier that could potentially provide more precise diagnosis and aid diagnosis in the near future.

KEY POINTS

• A precise preoperative prediction of the WHO grade of a meningioma brings benefits to further treatment plans. • Machine learning models based on clinical, morphological features and ADC value could achieve equivalent diagnostic performance compared to experienced neuroradiologists. • The decision forest model built with 23 selected texture features and the ADC value achieved the best diagnostic performance (kappa = 0.64, accuracy = 79.51%).