Differentiation of primary central nervous system lymphomas and glioblastomas: comparisons of diagnostic performance of dynamic susceptibility contrast-enhanced perfusion MR imaging without and with contrast-leakage correction

Imaging Features of Immunodeficiency-Associated Primary CNS Lymphoma: A Review

In the immunocompromised setting, there are distinct radiologic findings of primary central nervous system lymphoma (PCNSL), including necrotic ring-enhancing lesions, increased propensity for intralesional haemorrhage, and multiplicity. In this clinical context, advanced imaging with MR perfusion, spectroscopy, and diffusion-weighted imaging can be used to increase accuracy in the diagnosis of lymphoma over mimics such as high-grade glioma, metastases, or infection. This review summarizes the histology and pathophysiology of PCNSL in immunodeficient hosts, which provide a basis for its imaging appearances, prognosis, and treatment. This discussion is important for the general radiologist as the incidence of immunodeficiency-related PCNSL may be increasing.

An accessible deep learning tool for voxel-wise classification of brain malignancies from perfusion MRI

Noninvasive differential diagnosis of brain tumors is currently based on the assessment of magnetic resonance imaging (MRI) coupled with dynamic susceptibility contrast (DSC). However, a definitive diagnosis often requires neurosurgical interventions that compromise patients' quality of life. We apply deep learning on DSC images from histology-confirmed patients with glioblastoma, metastasis, or lymphoma. The convolutional neural network trained on ∼50,000 voxels from 40 patients provides intratumor probability maps that yield clinical-grade diagnosis. Performance is tested in 400 additional cases and an external validation cohort of 128 patients. The tool reaches a three-way accuracy of 0.78, superior to the conventional MRI metrics cerebral blood volume (0.55) and percentage of signal recovery (0.59), showing high value as a support diagnostic tool. Our open-access software, Diagnosis In Susceptibility Contrast Enhancing Regions for Neuro-oncology (DISCERN), demonstrates its potential in aiding medical decisions for brain tumor diagnosis using standard-of-care MRI.

Multiparametric analysis from dynamic susceptibility contrast-enhanced perfusion MRI to evaluate malignant brain tumors

BACKGROUND AND PURPOSE

Dynamic susceptibility contrast-enhanced (DSC) MR perfusion is a valuable technique for distinguishing brain tumors. Diagnostic potential of measurable parameters derived from preload leakage-corrected-DSC-MRI remains somewhat underexplored. This study aimed to evaluate these parameters for differentiating primary CNS lymphoma (PCNSL), glioblastoma, and metastasis.

METHODS

Thirty-nine patients with pathologically proven PCNSL (n = 14), glioblastoma (n = 14), and metastasis (n = 11) were analyzed. Five DSC parameters-relative CBV (rCBV), percentage of signal recovery (PSR), downward slope (DS), upward slope (US), and first-pass slope ratio-were derived from tumor-enhancing areas. Diagnostic performance was assessed using receiver operating characteristic curve analysis.

RESULTS

RCBV was higher in metastasis (4.58; interquartile range [IQR]: 2.54) and glioblastoma (3.98; IQR: 1.87), compared with PCNSL (1.46; IQR: 0.29; p = .00006 for both). rCBV better distinguished metastasis and glioblastoma from PCNSL, with an area under the curve (AUC) of 0.97 and 0.99, respectively. PSR was higher in PCNSL (88.11; IQR: 21.21) than metastases (58.30; IQR: 22.28; p = .0002), while glioblastoma (74.54; IQR: 21.23) presented almost significant trend-level differences compared to the others (p≈.05). AUCs were 0.79 (PCNSL vs. glioblastoma), 0.91 (PCNSL vs. metastasis), and 0.78 (glioblastoma vs. metastasis). DS and US parameters were statistically significant between glioblastoma (-109.92; IQR: 152.71 and 59.06; IQR: 52.87) and PCNSL (-47.36; IQR: 44.30 and 21.68; IQR: 16.85), presenting AUCs of 0.86 and 0.87.

CONCLUSION

Metastasis and glioblastoma can be better differentiated from PCNSL through rCBV. PSR demonstrated higher differential performance compared to the other parameters and seemed useful, allowing a proper distinction among all, particularly between metastasis and glioblastoma, where rCBV failed. Finally, DS and US were only helpful in differentiating glioblastoma from PCNSL.

Postoperative disappearance of leptomeningeal enhancement around the brainstem in glioblastoma

PURPOSE

Leptomeningeal enhancement (LME) suggests leptomeningeal dissemination (LMD) of tumor cells, which is a complication of end-stage glioblastoma, and is associated with a poor prognosis. However, magnetic resonance imaging (MRI) occasionally indicates the disappearance of peri-brainstem LME after surgical resection of glioblastoma. Since preoperative LMD may affect treatment indications, we aimed to analyze the clinical significance of preoperative LME of the brainstem in glioblastoma.

METHODS

We retrospectively collected clinical and radiological data from consecutive patients with glioblastoma and preoperative LME of the brainstem, who were treated at our hospital between 2017 and 2020.

RESULTS

Among 112 patients with glioblastoma, nine (8%) showed preoperative LME of the brainstem. In comparison with tumors without LME, tumor size was significantly associated with the preoperative LME of the brainstem (p = 0.016). In addition, there was a trend toward significance for a relationship between deep tumor location and preoperative LME of the brainstem (p = 0.058). Notably, among six patients who underwent surgical resection for glioblastoma with LME of the brainstem, four showed significant radiological disappearance of the LME on postoperative MRI. This suggests that the LME did not result from LMD in these cases. Moreover, these four patients lived longer than would be expected from the presence of LMD. However, this LME disappearance was not observed after biopsy or chemoradiotherapy.

CONCLUSIONS

These findings suggest that preoperative LME does not necessarily indicate the presence of untreatable LMD; moreover, LME may disappear after surgical tumor resection. Thus, transient preoperative LME could be attributed to other mechanisms, including impaired venous flow due to intratumoral arteriovenous shunts, which can be resolved by reducing the tumor burden.

Clinical Value of a Nomogram Model Based on Apparent Diffusion Coefficient Values Within 1 cm of the Tumor Cavity to Predict Postoperative Progression of Glioma

OBJECTIVE

To explore the clinical value of constructing a nomogram model based on apparent diffusion coefficient values within 1 cm of the residual tumor cavity to predict the postoperative progression of gliomas.

METHODS

Clinical data of patients with glioma who underwent surgery were retrospectively retrieved from the First Hospital of Qinhuangdao. The mean apparent diffusion coefficient (mADC) was measured using a picture archiving and communication system. The Kaplan-Meier survival curve was constructed with the optimal mADC threshold determined by the X-tile. A nomogram was developed based on the independent risk factors determined using the Cox proportional hazards model (Cox regression model) to predict the progression of postoperative glioma. A receiver operating characteristic curve was drawn to evaluate the prediction accuracy of the model, and decision curve analysis was performed to assess the clinical value of the nomogram.

RESULTS

There was good agreement between the mADC values of the 2 repeated measurements before and after, with a consistency correlation coefficient of 0.83. Multivariate Cox regression analysis showed that peritumoral mADC values, degree of peritumoral enhancement, age, pathological grading, and degree of tumor resection were independent risk factors for predicting postoperative progression of glioma (all P < 0.05). The receiver operating characteristic curves of the nomogram predicting 1, 2, and 3 years postoperative progression were 0.86, 0.82, and 0.91, respectively. The calibration curve showed good consistency between the observed and predicted values in the model. The curve showed that the nomogram model has a good clinical application value.

CONCLUSIONS

The peritumoral mADC values, degree of peritumoral enhancement, age, pathological grade, and degree of tumor resection were independent factors affecting the postoperative progression of glioma. The nomogram model established for the first time based on mADC values within 1 cm of the tumor can predict the postoperative condition of patients with glioma intuitively and comprehensively. It can provide a relatively accurate prediction tool for neurosurgeons to individualize the evaluation of survival and prognosis, and formulate treatment plans for patients.

Primary brain tumours in adults

The most frequent adult-type primary CNS tumours are diffuse gliomas, but a large variety of rarer CNS tumour types exists. The classification of these tumours is increasingly based on molecular diagnostics, which is reflected in the extensive molecular foundation of the recent WHO 2021 classification of CNS tumours. Resection as extensive as is safely possible is the cornerstone of treatment in most gliomas, and is now also recommended early in the treatment of patients with radiological evidence of histologically low-grade tumours. For the adult-type diffuse glioma, standard of care is a combination of radiotherapy and chemotherapy. Although treatment with curative intent is not available, combined modality treatment has resulted in long-term survival (>10-20 years) for some patients with isocitrate dehydrogenase (IDH) mutant tumours. Other rarer tumours require tailored approaches, best delivered in specialised centres. Targeted treatments based on molecular alterations still only play a minor role in the treatment landscape of adult-type diffuse glioma, and today are mainly limited to patients with tumours with BRAFV600E (ie, Val600Glu) mutations. Immunotherapy for CNS tumours is still in its infancy, and so far, trials with checkpoint inhibitors and vaccination studies have not shown improvement in patient outcomes in glioblastoma. Current research is focused on improving our understanding of the immunosuppressive tumour environment, the molecular heterogeneity of tumours, and the role of tumour microtube network connections between cells in the tumour microenvironment. These factors all appear to play a role in treatment resistance, and indicate that novel approaches are needed to further improve outcomes of patients with CNS tumours.

Distinctive magnetic resonance imaging features in primary central nervous system lymphoma: A case report

BACKGROUND

Primary central nervous system lymphoma (PCNSL) is a rare malignant tumor originating from the lymphatic hematopoietic system. It exhibits unique imaging manifestations due to its biological characteristics.

CASE SUMMARY

Magnetic resonance imaging (MRI) with diffusion-weighted imaging (DWI), perfusion-weighted imaging (PWI), and magnetic resonance spectroscopy was performed. The imaging findings showed multiple space-occupying lesions with low signal on T1-weighted imaging, uniform high signal on T2-weighted imaging, and obvious enhancement on contrast-enhanced scans. DWI revealed diffusion restriction, PWI demonstrated hypoperfusion, and spectroscopy showed elevated choline peak and decreased N-acetylaspartic acid. The patient's condition significantly improved after hormone shock therapy.

CONCLUSION

This case highlights the distinctive imaging features of PCNSL and their importance in accurate diagnosis and management.

Machine learning decision tree models for multiclass classification of common malignant brain tumors using perfusion and spectroscopy MRI data

Background

To investigate the contribution of machine learning decision tree models applied to perfusion and spectroscopy MRI for multiclass classification of lymphomas, glioblastomas, and metastases, and then to bring out the underlying key pathophysiological processes involved in the hierarchization of the decision-making algorithms of the models.

Methods

From 2013 to 2020, 180 consecutive patients with histopathologically proved lymphomas (n = 77), glioblastomas (n = 45), and metastases (n = 58) were included in machine learning analysis after undergoing MRI. The perfusion parameters (rCBVmax, PSRmax) and spectroscopic concentration ratios (lac/Cr, Cho/NAA, Cho/Cr, and lip/Cr) were applied to construct Classification and Regression Tree (CART) models for multiclass classification of these brain tumors. A 5-fold random cross validation was performed on the dataset.

Results

The decision tree model thus constructed successfully classified all 3 tumor types with a performance (AUC) of 0.98 for PCNSLs, 0.98 for GBM and 1.00 for METs. The model accuracy was 0.96 with a RSquare of 0.887. Five rules of classifier combinations were extracted with a predicted probability from 0.907 to 0.989 for that end nodes of the decision tree for tumor multiclass classification. In hierarchical order of importance, the root node (Cho/NAA) in the decision tree algorithm was primarily based on the proliferative, infiltrative, and neuronal destructive characteristics of the tumor, the internal node (PSRmax), on tumor tissue capillary permeability characteristics, and the end node (Lac/Cr or Cho/Cr), on tumor energy glycolytic (Warburg effect), or on membrane lipid tumor metabolism.

Conclusion

Our study shows potential implementation of machine learning decision tree model algorithms based on a hierarchical, convenient, and personalized use of perfusion and spectroscopy MRI data for multiclass classification of these brain tumors.

Atypical primary central nervous system lymphoma and glioblastoma: multiparametric differentiation based on non-enhancing volume, apparent diffusion coefficient, and arterial spin labeling

OBJECTIVES

To evaluate the multiparametric diagnostic performance with non-enhancing tumor volume, apparent diffusion coefficient (ADC), and arterial spin labeling (ASL) to differentiate between atypical primary central nervous system lymphoma (PCNSL) and glioblastoma (GBM).

METHODS

One hundred and fifty-eight patients with pathologically confirmed typical PCNSL (n = 59), atypical PCNSL (hemorrhage, necrosis, or heterogeneous contrast enhancement, n = 29), and GBM (n = 70) were selected. Relative minimum ADC (rADCmin), mean (rADCmean), maximum (rADCmax), and rADCmax-min (rADCdif) were obtained by standardization of the contralateral white matter. Maximum cerebral blood flow (CBFmax) was obtained according to the ASL-CBF map. The regions of interests (ROIs) were manually delineated on the inner side of the tumor to further generate a 3D-ROI and obtain the non-enhancing tumor (nET) volume. The area under the curve (AUC) was used to evaluate the diagnostic performance.

RESULTS

Atypical PCNSLs showed significantly lower rADCmax, rADCmean, and rADCdif than that of GBMs. GBMs showed significantly higher CBFmax and nET volume ratios than that of atypical PCNSLs. Combined three-variable models with rADCmean, CBFmax, and nET volume ratio were superior to one- and two-variable models. The AUC of the three-variable model was 0.96, and the sensitivity and specificity were 90% and 96.55%, respectively.

CONCLUSION

The combined evaluation of rADCmean, CBFmax, and nET volume allowed for reliable differentiation between atypical PCNSL and GBM.

KEY POINTS

• Atypical PCNSL is easily misdiagnosed as glioblastoma, which leads to unnecessary surgical resection. • The nET volume, ADC, and ASL-derived parameter (CBF) were lower for atypical PCNSL than that for glioblastoma. • The combination of multiple parameters performed well (AUC = 0.96) in the discrimination between atypical PCNSL and glioblastoma.

Evaluation of the Diagnostic Performance of F18-Fluorodeoxyglucose-Positron Emission Tomography, Dynamic Susceptibility Contrast Perfusion, and Apparent Diffusion Coefficient in Differentiation between Recurrence of a High-grade Glioma and Radiation Necrosis

Background

Differentiation between recurrence of brain tumor and radiation necrosis remains a challenge in current neuro-oncology practice despite recent advances in both radiological and nuclear medicine techniques.

Purpose

The purpose of this study was to compare the diagnostic performance of dynamic susceptibility contrast (DSC) perfusion magnetic resonance imaging (MRI), apparent diffusion coefficient (ADC) derived from diffusion-weighted imaging, and F18-fluorodeoxyglucose-positron emission tomography (F18-FDG-PET) in the differentiation between the recurrence of a high-grade glioma and radiation necrosis.

Materials and Methods

Patients with a diagnosis of high-grade glioma (WHO Grades III and IV) who had undergone surgical resection of the tumor followed by radiotherapy with or without chemotherapy were included in the study. DSC perfusion, diffusion-weighted MRI, and PET scan were acquired on a hybrid PET/MRI scanner. For each lesion, early and delayed tumor-to-brain ratio (TBR), early and delayed maximum standardized uptake value (SUVmax), normalized ADC ratio, and normalized relative cerebral blood volume (rCBV) ratio were calculated and the pattern of lesional enhancement was noted. The diagnosis was finalized with either histopathological examination or the characteristics on follow-up imaging. The statistical analysis using the receiver operator characteristic curves was done to determine the diagnostic performance of DSC perfusion, 18-F FDG-PET, and ADC in differentiation between tumor recurrence and radiation necrosis.

Results

Fifty patients were included in the final analysis, 32 of them being men (64%). A cutoff value of early TBR >0.8 (sensitivity of 100% and specificity of 80%), delayed TBR >0.93 (sensitivity of 92.3% and specificity of 80%), early SUVmax >10.2 (sensitivity of 76.9% and specificity of 80%), delayed SUVmax >13.2 (sensitivity of 61.54% and specificity of 100%), normalized rCBV ratio >1.21 (sensitivity of 100% and specificity of 60%), normalized ADC ratio >1.66 (sensitivity of 38.5% and specificity of 80%), and Grade 3 enhancement (sensitivity of 100% and specificity of 60%) were found to differentiate recurrence from radiation necrosis. Early TBR had the highest accuracy (94.44%), while ADC ratio had the lowest accuracy (50%). A combination of early TBR (cutoff value of 0.8), late TBR (cutoff value of 0.93), and rCBV ratio (cutoff value of 1.21) showed a sensitivity of 100%, specificity of 92.3%, positive predictive value of 88.9%, negative predictive value of 93.7%, and an accuracy of 96.6% in discrimination between radiation necrosis and recurrence of tumor.

Conclusion

F18-FDG-PET and DSC perfusion can reliably differentiate tumor recurrence from radiation necrosis, with early TBR showing the highest accuracy. ADC demonstrates a low sensitivity, specificity, and accuracy in differentiating radiation necrosis from recurrence. A combination of early TBR, delayed TBR, and rCBV may be more useful in discrimination between radiation necrosis and recurrence of glioma, with this combination showing a better diagnostic performance than individual parameters or any other combination of parameters.

Hemodynamic Imaging in Cerebral Diffuse Glioma-Part A: Concept, Differential Diagnosis and Tumor Grading

Diffuse gliomas are the most common primary malignant intracranial neoplasms. Aside from the challenges pertaining to their treatment-glioblastomas, in particular, have a dismal prognosis and are currently incurable-their pre-operative assessment using standard neuroimaging has several drawbacks, including broad differentials diagnosis, imprecise characterization of tumor subtype and definition of its infiltration in the surrounding brain parenchyma for accurate resection planning. As the pathophysiological alterations of tumor tissue are tightly linked to an aberrant vascularization, advanced hemodynamic imaging, in addition to other innovative approaches, has attracted considerable interest as a means to improve diffuse glioma characterization. In the present part A of our two-review series, the fundamental concepts, techniques and parameters of hemodynamic imaging are discussed in conjunction with their potential role in the differential diagnosis and grading of diffuse gliomas. In particular, recent evidence on dynamic susceptibility contrast, dynamic contrast-enhanced and arterial spin labeling magnetic resonance imaging are reviewed together with perfusion-computed tomography. While these techniques have provided encouraging results in terms of their sensitivity and specificity, the limitations deriving from a lack of standardized acquisition and processing have prevented their widespread clinical adoption, with current efforts aimed at overcoming the existing barriers.

3D pseudo-continuous arterial spin labeling-MRI (3D PCASL-MRI) in the differential diagnosis between glioblastomas and primary central nervous system lymphomas

PURPOSE

The aim of the study was to compare the parameters of blood flow in glioblastomas and primary central nervous system lymphomas (PCNSLs), measured by pseudo-continuous arterial spin labeling MRI (3D PCASL), and to determine the informativeness of this method in the differential diagnosis between these lesions.

METHODS

The study included MRI data of 139 patients with PCNSL (n = 21) and glioblastomas (n = 118), performed in the Burdenko Neurosurgical Center. No patients received chemotherapy, hormone therapy, or radiation therapy prior to MRI. On the 3D PCASL perfusion map, the absolute and normalized values of tumor blood flow were calculated in the glioblastoma and PCNSL groups (maxTBFmean and nTBF).

RESULTS

MaxTBFmean and nTBF in the glioblastoma group were significantly higher than those in the PCNSL group: 168.9 ml/100 g/min versus 65.6 and 9.3 versus 3.7, respectively (p < 0.001). Arterial spin labeling perfusion had high sensitivity (86% for maxTBFmean, 95% for nTBF) and specificity (77% for maxTBFmean, 73% for nTBF) in the differential diagnosis between PCNSL and glioblastomas. Blood flow thresholds were 98.9 ml/100 g/min using absolute blood flow values and 6.1 using normalized values, AUC > 0.88.

CONCLUSION

The inclusion of 3D PCASL in the standard MRI protocol can increase the specificity of the differential diagnosis between glioblastomas and PCNSL.

Dynamic contrast-enhanced (DCE) imaging: state of the art and applications in whole-body imaging

Dynamic contrast-enhanced (DCE) imaging is a non-invasive technique used for the evaluation of tissue vascularity features through imaging series acquisition after contrast medium administration. Over the years, the study technique and protocols have evolved, seeing a growing application of this method across different imaging modalities for the study of almost all body districts. The main and most consolidated current applications concern MRI imaging for the study of tumors, but an increasing number of studies are evaluating the use of this technique also for inflammatory pathologies and functional studies. Furthermore, the recent advent of artificial intelligence techniques is opening up a vast scenario for the analysis of quantitative information deriving from DCE. The purpose of this article is to provide a comprehensive update on the techniques, protocols, and clinical applications - both established and emerging - of DCE in whole-body imaging.

Differentiation between primary CNS lymphoma and atypical glioblastoma according to major genomic alterations using diffusion and susceptibility-weighted MR imaging

PURPOSE

We aimed to differentiate primary central nervous system lymphoma (PCNSL) from atypical glioblastoma (GB) and distinguish major genomic subtypes between these tumors using susceptibility-weighted imaging (SWI) along with diffusion-weighted imaging (DWI).

METHODS

Thirty-one immuno-competent patients with PCNSL stratified by BCL2 and MYC rearrangement, and 57 patients with atypical GB (no visible necrosis) grouped according to isocitrate dehydrogenase-1 (IDH1) mutation status underwent 3.0-Tesla MRI before treatment in this retrospective study. Region of interest analysis with apparent diffusion coefficient (ADC) and SWI signal intensity values of the tumors were normalized by dividing those of contralateral white matter. The independent-samples t-test and Kruskal-Wallis test were utilized to compare parameters. The diagnostic ability of each parameter and their optimal combination was evaluated by logistic regression analysis and receiver operating characteristic.

RESULTS

PCNSL with rearrangement of both MYC and BCL2 (n = 7) [mean relative (r) ADCmean:0.87 ± 0.06, rADCmin:0.72 ± 0.08] demonstrated significantly lower rADCmean, and rADCmin compared to other PCNSLs (n = 24) (rADCmean:1.19 ± 0.18, rADCmin:1.03 ± 0.17;p < 0.001) and GBs (p < 0.001). GB without IDH1 mutation (n = 44) (mean rSWI value:0.95 ± 0.15) demonstrated significantly lower rSWI value compared to GB with IDH1 mutation (n = 13) (rSWI value:1.13 ± 0.09;p < 0.001) and PCNSL (p < 0.001). The incorporation of rADCmean and rSWI parameters distinguished GB with IDH1 mutation [Area under the curve (AUC):0.985] with sensitivity and specificity of 94.3 and 100 % respectively; and PCNSL with rearrangement of both MYC and BCL2 (AUC:0.982) with sensitivity and specificity of 100 % and 95.4 %, respectively. CONCLUSıONS: Combined analysis of SWI and DWI could differentiate atypical GB from PCNSL and distinguish major genomic subtypes between these tumors.

Differentiation of Primary Central Nervous System Lymphoma and High-Grade Glioma with Dynamic Susceptibility Contrast-Derived Metrics: Pilot Study

OBJECTIVE

Preoperative differentiation of lymphoma from other aggressive intracranial neoplasms is important as the surgical and adjuvant therapy may be fundamentally different between the 2 types of tumors. The purpose of this study was to assess the ability of the dynamic susceptibility contrast-derived metrics, percentage signal recovery (PSR) ratio, and relative cerebral blood volume (rCBV) to distinguish between primary central nervous system lymphoma (PCNSL) and high-grade glioma (HGG).

METHODS

Twenty-six patients (15 with HGG and 11 with PCNSL) with histologically confirmed diagnoses were retrospectively analyzed. Mean PSR and rCBV were calculated from dynamic susceptibility contrast imaging. The 2 groups were compared using an independent samples t-test. Receiver operating characteristic analyses were performed to determine the area under the curve and identify threshold values to differentiate PCNSL from GBM.

RESULTS

Both rCBV and PSR values were significantly different, at both the group level and subject level, between the PCNSL and HGG patients. The mean rCBV was significantly lower in PCNSL (1.38 ± 0.64) compared with HGG (5.19 ± 2.21, df = 11.24, P < 0.001). The mean PSR ratio was significantly higher in PCNSL (1.04 ± 0.11) compared with HGG (0.72 ± 0.16, df = 17.23, P < 0.001). An rCBV threshold value of 2.67 provided a 100% sensitivity and 100% specificity (area under the curve 1.0) for differentiating PCNSL from HGG. A PSR ratio threshold value of 0.9 was 100% sensitive and 90.91% specific for differentiating PCNSL from HGG.

CONCLUSIONS

The findings of our study show that rCBV and PSR ratio are different in HGG and PCNSL at both the group level and subject level. Incorporation of perfusion in routine magnetic resonance imaging of contrast-enhancing lesions can have a significant impact on patient management.

Differentiation between glioblastoma and primary CNS lymphoma: application of DCE-MRI parameters based on arterial input function obtained from DSC-MRI

OBJECTIVE

This study aimed to evaluate whether arterial input functions (AIFs) obtained from dynamic susceptibility contrast (DSC)-MRI (AIF_DSC) improve the reliability and diagnostic accuracy of dynamic contrast-enhanced (DCE)-derived pharmacokinetic (PK) parameters for differentiating glioblastoma from primary CNS lymphoma (PCNSL) compared with AIFs derived from DCE-MRI (AIF_DCE).

METHODS

This retrospective study included 172 patients with glioblastoma (n = 147) and PCNSL (n = 25). All patients had undergone preoperative DSC- and DCE-MRI. The volume transfer constant (K^trans), volume of the vascular plasma space (v_p), and volume of the extravascular extracellular space (v_e) were acquired using AIF_DSC and AIF_DCE. The relative cerebral blood volume (rCBV) was obtained from DSC-MRI. Intraclass correlation coefficients (ICC) and ROC curves were used to assess the reliability and diagnostic accuracy of individual parameters.

RESULTS

The mean K^trans, v_p, and v_e values revealed better ICCs with AIF_DSC than with AIF_DCE (K^trans, 0.911 vs 0.355; v_p, 0.766 vs 0.503; v_e, 0.758 vs 0.657, respectively). For differentiating all glioblastomas from PCNSL, the mean rCBV (AUC = 0.856) was more accurate than the AIF_DSC-driven mean K^trans, which had the largest AUC (0.711) among the DCE-derived parameters (p = 0.02). However, for glioblastomas with low rCBV (≤ 75th percentile of PCNSL; n = 30), the AIF_DSC-driven mean K^trans and v_p were more accurate than rCBV (AUC: K^trans, 0.807 vs rCBV, 0.515, p = 0.004; v_p, 0.715 vs rCBV, p = 0.045).

CONCLUSION

DCE-derived PK parameters using the AIF_DSC showed improved reliability and diagnostic accuracy for differentiating glioblastoma with low rCBV from PCNSL.

KEY POINTS

• An accurate differential diagnosis of glioblastoma and PCNSL is crucial because of different therapeutic strategies. • In contrast to the rCBV from DSC-MRI, another perfusion imaging technique, the DCE parameters for the differential diagnosis have been limited because of the low reliability of AIFs from DCE-MRI. • When we analyzed DCE-MRI data using AIFs from DSC-MRI (AIF_DSC), AIF_DSC-driven DCE parameters showed improved reliability and better diagnostic accuracy than rCBV for differentiating glioblastoma with low rCBV from PCNSL.

Peritumoral brain edema volume in meningioma correlates with tumor fractional anisotropy but not apparent diffusion coefficient or cerebral blood volume

PURPOSE

Pathogenesis of peritumoral brain edema (PTBE) in meningiomas remains unclear. Associations between PTBE volume and diffusion or perfusion properties of meningioma have not been studied. We aimed to investigate if diffusion and perfusion properties of meningioma correlate with its PTBE volume.

METHODS

Seventy consecutive patients (mean age, 58.9 ± 13.7 years; 37 women) with meningiomas who had preoperative DTI and DSC-PWI were retrospectively analyzed. PTBE volume, tumor volume, and mean T2 signal, ADC, FA, and CBV of the tumor were measured. Between meningiomas with and without PTBE, patient age and sex, as well as T2 signal intensity, volume, ADC, FA, and CBV of tumors, were compared. In meningiomas with PTBE, correlations of PTBE volume with patient age and sex, as well as T2 signal intensity, volume, ADC, FA, and CBV of tumors, were analyzed. Multivariable linear regression analysis was performed to identify factors associated with PTBE volume.

RESULTS

On univariable analysis, meningiomas without PTBE were more frequently found in women (P = 0.033) and demonstrated lower ADC (P = 0.020), higher FA (P < 0.001), and lower CBV (P < 0.001). PTBE volume of meningiomas correlated with tumor ADC (r = 0.444; P = 0.001), tumor FA (r = - 0.655; P < 0.001), and tumor CBV (r = 0.402; P = 0.003). On multivariable analysis, tumor FA was the only factor associated with PTBE volume (P < 0.001).

CONCLUSION

PTBE volume in meningioma correlates with tumor FA. DTI may help to understand the mechanism of PTBE in meningiomas.

Differentiating Glioblastoma from Primary Central Nervous System Lymphoma: The Value of Shaping and Nonenhancing Peritumoral Hyperintense Gyral Lesion on FLAIR Imaging

BACKGROUND

This study describes a distinct magnetic resonance imaging (MRI) feature, placing emphasis on fluid-attenuation inversion recovery (FLAIR) and contrast-enhanced T1-weighted (T1C) images for the preoperative differentiation of glioblastoma (GBM) from primary central nervous system lymphoma (PCNSL).

METHODS

The preoperative MRI findings of 116 pathologically confirmed glioblastoma (n = 72) and PCNSL (n = 44) were retrospectively reviewed. Two neuroimaging specialists analyzed the MRIs, and image analysis was focused on the presence or absence of a shaping and nonenhancing peritumoral hyperintense gyral lesion on FLAIR imaging (SNEPGF, i.e., hyperintense lesion in a shaping and nonenhancing peritumoral gyral area on FLAIR imaging). The gyral area adjacent to and within 3 cm of the enhanced tumor was defined as the peritumoral gyrus region. The FLAIR hyperintensity lesion were termed as the signal intensity ratio ≥30% compared with contralateral normal gray matter. Then, the differential diagnostic efficacy of SNEFPG sign for GBM and PCNSL was analyzed.

RESULTS

The SNEPGF sign was found in 33 GBM cases (33 of 72, 45.8%), and the FLAIR signal intensity and apparent diffusion coefficient value of these area were lower than the peritumoral edema area (P < 0.0001). In 44 PCNSL cases, no SNEPGF sign was found. A slightly higher FLAIR signal intensity was seen in 9 PCNSLs, but uniform and marked enhancement was seen in these areas. The sensitivity, specificity, positive predictive value, and negative predictive value of the differential diagnosis of GBM and PCNSL with SNEPGF sign were 45.8%, 100%, 100%, and 53.0%, respectively.

CONCLUSIONS

The SNEPGF sign is effective in identifying GBM from PCNSL, especially with high specificity.

Classification of Primary Cerebral Lymphoma and Glioblastoma Featuring Dynamic Susceptibility Contrast and Apparent Diffusion Coefficient

This study aimed to differentiate primary central nervous system lymphoma (PCNSL) and glioblastoma (GBM) via multimodal MRI featuring radiomic analysis. MRI data sets of patients with histological proven PCNSL and GBM were analyzed retrospectively. Diffusion-weighted imaging (DWI) and dynamic susceptibility contrast (DSC) perfusion imaging were evaluated to differentiate contrast enhancing intracerebral lesions. Selective (contrast enhanced tumor area with the highest mean cerebral blood volume (CBV) value) and unselective (contouring whole contrast enhanced lesion) Apparent diffusion coefficient (ADC) measurement was performed. By multivariate logistic regression, a multiparametric model was compiled and tested for its diagnostic strength. A total of 74 patients were included in our study. Selective and unselective mean and maximum ADC values, mean and maximum CBV and ratio_CBV as quotient of tumor CBV and CBV in contralateral healthy white matter were significantly larger in patients with GBM than PCNSL; minimum CBV was significantly lower in GBM than in PCNSL. The highest AUC for discrimination of PCNSL and GBM was obtained for selective mean and maximum ADC, mean and maximum CBV and ratio_CBV. By integrating these five in a multiparametric model 100% of the patients were classified correctly. The combination of perfusion imaging (CBV) and tumor hot-spot selective ADC measurement yields reliable radiological discrimination of PCNSL from GBM with highest accuracy and is readily available in clinical routine.

Percentage signal intensity recovery: A step ahead of rCBV in DSC MR perfusion imaging for the differentiation of common neoplasms of brain

Context

Relative cerebral blood volume (rCBV) and percentage signal recovery (PSR) obtained from T2* dynamic susceptibility contrast magnetic resonance imaging are important parameters for brain tumor assessment.

Aim

To study the accuracy of PSR in the differentiation of low-grade glioma, high-grade glioma, lymphoma, and metastases particularly in comparison to rCBV.

Settings and Design

Retrospective observational study.

Subjects and Methods

Study included pathologically confirmed cases of 10 low-grade glioma, 22 high-grade glioma, 6 lymphoma, and 12 metastases (Total 50). PSR, relative PSR (rPSR), and rCBV were calculated.

Statistical Analysis Used

Accuracy of these parameters studied statistically using analysis of variance and ROC (Receiver operating characteristic) curves.

Results

rCBV was higher in metastases (3.45 ± 2.82) and high-grade glioma (3.47 ± 1.62), whereas was low in lymphoma (1.03 ± 0.74) and low-grade glioma (1.43 ± 0.47) with P value of 0.030. PSR was low in metastases (48 ± 16.18), intermediate in glioma (73.24 ± 6.39 and 88.26 ± 6.05, high and low grade), and high in lymphoma (112.16 ± 10.57) with P value < 0.000. rPSR was higher for lymphoma (1.73 ± 0.57) than high-grade glioma (0.85 ± 0.11) and metastasis (0.69 ± 0.19) with P value <.000. Area under ROC for PSR was greater than rCBV in differentiating metastases from lymphoma (1.00 vs 0.13), high-grade glioma from lymphoma (1.00 vs 0.38), high-grade glioma from metastases (0.89 vs 0.58), and high-grade glioma from low-grade glioma (0.96 vs 0.03) with excellent curve characteristics. F values for PSR and rPSR from ANOVA analysis were 71.47 and 36.77, was better than rCBV (3.84) in differentiating these groups.

Conclusions

Percentage of signal recovery shows low recovery values in metastases, intermediate recovery values in glioma, and overshoot in lymphoma. PSR values show lower overlap than rCBV between lymphoma and metastases; and between high grade glioma and metastases. PSR difference is also higher than rCBV between low- and high-grade gliomas. Hence, PSR can potentially help as an additional perfusion parameter in the preoperative differentiation of these tumors.

Comparison of Radiomics-Based Machine-Learning Classifiers in Diagnosis of Glioblastoma From Primary Central Nervous System Lymphoma

Purpose: The purpose of the current study was to evaluate the ability of magnetic resonance (MR) radiomics-based machine-learning algorithms in differentiating glioblastoma (GBM) from primary central nervous system lymphoma (PCNSL).

Method: One-hundred and thirty-eight patients were enrolled in this study. Radiomics features were extracted from contrast-enhanced MR images, and the machine-learning models were established using five selection methods (distance correlation, random forest, least absolute shrinkage and selection operator (LASSO), eXtreme gradient boosting (Xgboost), and Gradient Boosting Decision Tree) and three radiomics-based machine-learning classifiers [linear discriminant analysis (LDA), support vector machine (SVM), and logistic regression (LR)]. Sensitivity, specificity, accuracy, and areas under curves (AUC) of models were calculated, with which the performances of classifiers were evaluated and compared with each other.

Result: Brilliant discriminative performance would be observed among all classifiers when combined with the suitable selection method. For LDA-based models, the optimal one was Distance Correlation + LDA with AUC of 0.978. For SVM-based models, Distance Correlation + SVM was the one with highest AUC of 0.959, while for LR-based models, the highest AUC was 0.966 established with LASSO + LR.

Conclusion: Radiomics-based machine-learning algorithms potentially have promising performances in differentiating GBM from PCNSL.

Validation of combined use of DWI and percentage signal recovery-optimized protocol of DSC-MRI in differentiation of high-grade glioma, metastasis, and lymphoma

PURPOSE

With conventional MRI, it is often difficult to effectively differentiate between contrast-enhancing brain tumors, including primary central nervous system lymphoma (PCNSL), high-grade glioma (HGG), and metastasis. This study aimed to assess the discrimination ability of the parameters obtained from DWI and the percentage signal recovery- (PSR-) optimized protocol of DSC-MRI between these three tumor types at an initial step.

METHODS

DSC-MRI using a PSR-optimized protocol (TR/TE = 1500/30 ms, flip angle = 90°, no preload) and DWI of 99 solitary enhancing tumors (60 HGGs, 24 metastases, 15 PCNSLs) were retrospectively assessed before treatment. rCBV, PSR, ADC in the tumor core and rCBV, and ADC in peritumoral edema were measured. The differences were evaluated using one-way ANOVA, and the diagnostic performance was evaluated using ROC curve analysis.

RESULTS

PSR in the tumor core showed the best discriminating performance in differentiating these three tumor types with AUC values of 0.979 for PCNSL vs. others and 0.947 for HGG vs. metastasis. The ADC was only helpful in the tumor core and distinguishing PCNSLs from others (AUC = 0.897).

CONCLUSION

Different from CBV-optimized protocols (preload, intermediate FA), PSR derived from the PSR-optimized protocol seems to be the most important parameter in the differentiation of HGGs, metastases, and PCNSLs at initial diagnosis. This property makes PSR remarkable and carries the need for comprehensive DSC-MRI protocols, which provides PSR sensitivity and CBV accuracy together, such as the preload use of the PSR-optimized protocol before the CBV-optimized protocol.

Performance of diffusion and perfusion MRI in evaluating primary central nervous system lymphomas of different locations

Background

Diffusion and perfusion MRI can invasively define physical properties and angiogenic features of tumors, and guide the individual treatment. The purpose of this study was to investigate whether the diffusion and perfusion MRI parameters of primary central nervous system lymphomas (PCNSLs) are related to the tumor locations.

Methods

We retrospectively reviewed the diffusion, perfusion, and conventional MRI of 68 patients with PCNSLs at different locations (group 1: cortical gray matter, group 2: white matter, group 3: deep gray matter). Relative maximum cerebral blood volume (rCBV_max) from perfusion MRI, minimum apparent diffusion coefficients (ADC_min) from DWI of each group were calculated and compared by one-way ANOVA test. In addition, we compared the mean apparent diffusion coefficients (ADC_mean) in three different regions of control group.

Results

The rCBV_max of PCNSLs yielded the lowest value in the white matter group, and the highest value in the cortical gray matter group (P < 0.001). However, the ADC_min of each subgroup was not statistically different. The ADC_mean of each subgroup in control group was not statistically different.

Conclusion

Our study confirms that rCBV_max of PCNSLs are related to the tumor location, and provide simple but effective information for guiding the clinical practice of PCNSLs.

Correlation and Characteristics of Intravoxel Incoherent Motion and Arterial Spin Labeling Techniques Versus Multiple Parameters Obtained on Dynamic Susceptibility Contrast Perfusion MRI for Brain Tumors

Purpose : To compare data on brain tumors derived from intravoxel incoherent motion (IVIM) and arterial spin labeling (ASL) imaging with multiple parameters obtained on dynamic susceptibility contrast (DSC) perfusion MRI and to clarify the characteristics of IVIM and ASL perfusion data from the viewpoint of cerebral blood flow (CBF) analysis. Methods : ASL-CBF and IVIM techniques as well as DSC examination were performed in 24 patients with brain tumors. The IVIM data were analyzed with the two models. The relative blood flow (rBF), relative blood volume (rBV) corrected relative blood volume (crBV), mean transit time (MTT), and leakage coefficient (K2) were obtained from the DSC MRI data. Results : The ASL-CBF had the same tendency as the perfusion parameters derived from the DSC data, but the permeability from the vessels had less of an effect on the ASL-CBF. The diffusion coefficient of the fast component on IVIM contained more information on permeability than the f value. Conclusion : ASL-CBF is more suitable for the evaluation of perfusion in brain tumors than IVIM parameters. ASL-CBF and IVIM techniques should be carefully selected and the biological significance of each parameter should be understood for the correct comprehension of the pathological status of brain tumors. J. Med. Invest. 66 : 308-313, August, 2019.

Can we predict intraoperative blood loss in meningioma patients? Application of dynamic susceptibility contrast-enhanced magnetic resonance imaging

PURPOSE

To evaluate the potential of quantitative dynamic susceptibility contrast (DSC) perfusion MR imaging parameters as imaging biomarkers for predicting intraoperative blood loss in meningioma.

METHODS

Fifty-one non-embolized meningioma patients who had undergone preoperative DSC perfusion MR imaging were retrospectively included. The corrected relative cerebral blood volume (rCBV) and leakage coefficient (K₂) of the entire enhanced tumor were obtained using leakage correction. Tumor volume, location, grade, and other clinical variables, were also analyzed. To investigate the vascularity and vascular permeability of meningiomas, and their correlation with predicting estimated blood loss (EBL) using preoperative DSC perfusion MR imaging, the authors proposed an index reflecting the inherent tendency of meningiomas to bleed after controlling volume (i.e., EBL/cm³). Simple regression was performed to identify predictors of EBL/cm³; subsequently, the relevant variables included in the stepwise multiple linear regression.

RESULTS

On univariate analysis, EBL/cm³ was correlated with rCBV (r=0.677; P<0.001), K₂ (r=0.294; P=0.036), and tumor volume (r=-0.312, P=0.026). EBL/cm³ was not correlated with age (P=0.873), sex (P=0.404), tumor location (P=0.327), or histological grade (P=0.230). On multiple linear regression, rCBV (β=0.663 [0.463-0.864], B=1.293 [0.903-1.684; P<0.001) and K₂ (β=0.260 [0.060-0.460], B=2.277 [0.523-4.031], P=0.012), were the only independent predictors of EBL/cm³.

CONCLUSION

The rCBV and K₂ derived from DSC perfusion MR imaging in meningiomas may serve as feasible tools for clinicians to predict intraoperative blood loss and facilitate surgical planning.

Machine learning for semi-automated classification of glioblastoma, brain metastasis and central nervous system lymphoma using magnetic resonance advanced imaging

Background

Differentiating glioblastoma, brain metastasis, and central nervous system lymphoma (CNSL) on conventional magnetic resonance imaging (MRI) can present a diagnostic dilemma due to the potential for overlapping imaging features. We investigate whether machine learning evaluation of multimodal MRI can reliably differentiate these entities.

Methods

Preoperative brain MRI including diffusion weighted imaging (DWI), dynamic contrast enhanced (DCE), and dynamic susceptibility contrast (DSC) perfusion in patients with glioblastoma, lymphoma, or metastasis were retrospectively reviewed. Perfusion maps (rCBV, rCBF), permeability maps (K-trans, Kep, Vp, Ve), ADC, T1C+ and T2/FLAIR images were coregistered and two separate volumes of interest (VOIs) were obtained from the enhancing tumor and non-enhancing T2 hyperintense (NET2) regions. The tumor volumes obtained from these VOIs were utilized for supervised training of support vector classifier (SVC) and multilayer perceptron (MLP) models. Validation of the trained models was performed on unlabeled cases using the leave-one-subject-out method. Head-to-head and multiclass models were created. Accuracies of the multiclass models were compared against two human interpreters reviewing conventional and diffusion-weighted MR images.

Results

Twenty-six patients enrolled with histopathologically-proven glioblastoma (n=9), metastasis (n=9), and CNS lymphoma (n=8) were included. The trained multiclass ML models discriminated the three pathologic classes with a maximum accuracy of 69.2% accuracy (18 out of 26; kappa 0.540, P=0.01) using an MLP trained with the VpNET2 tumor volumes. Human readers achieved 65.4% (17 out of 26) and 80.8% (21 out of 26) accuracies, respectively. Using the MLP VpNET2 model as a computer-aided diagnosis (CADx) for cases in which the human reviewers disagreed with each other on the diagnosis resulted in correct diagnoses in 5 (19.2%) additional cases.

Conclusions

Our trained multiclass MLP using VpNET2 can differentiate glioblastoma, brain metastasis, and CNS lymphoma with modest diagnostic accuracy and provides approximately 19% increase in diagnostic yield when added to routine human interpretation.

A Case of Sarcoidosis-lymphoma Syndrome: Importance of Brain Biopsy

Sarcoidosis is occasionally associated with malignant lymphoma. Cases of sarcoidosis associated with malignant lymphoma are called sarcoidosis-lymphoma syndrome. We report a 63-year-old man with sarcoidosis associated with primary central nervous system lymphoma (PCNSL). Definitive diagnosis by clinical and radiological findings was difficult. Finally, brain biopsy could produce the correct diagnosis. We could provide appropriate treatment for PCNSL and the patient has survived over 2 years. Thus, it is very important to make an early definitive diagnosis by biopsy for intracranial lesion, because it can prolong survival in patients with sarcoidosis.

MRI as a diagnostic biomarker for differentiating primary central nervous system lymphoma from glioblastoma: A systematic review and meta-analysis

BACKGROUND

Accurate preoperative differentiation of primary central nervous system lymphoma (PCNSL) and glioblastoma is clinically crucial because the treatment strategies differ substantially.

PURPOSE

To evaluate the diagnostic performance of MRI for differentiating PCNSL from glioblastoma.

STUDY TYPE

Systematic review and meta-analysis.

SUBJECTS

Ovid-MEDLINE and EMBASE databases were searched to find relevant original articles up to November 25, 2018. The search term combined synonyms for "lymphoma," "glioblastoma," and "MRI."

FIELD STRENGTH/SEQUENCE

Patients underwent at least one MRI sequence including diffusion-weighted imaging (DWI), dynamic susceptibility-weighted contrast-enhanced imaging (DSC), dynamic contrast-enhanced imaging (DCE), arterial spin labeling (ASL), susceptibility-weighted imaging (SWI), intravoxel incoherent motion (IVIM), and magnetic resonance spectroscopy (MRS) using 1.5 or 3 T.

ASSESSMENT

Quality assessment was performed according to the Quality Assessment of Diagnostic Accuracy Studies-2 tool.

STATISTICAL TESTS

Hierarchical logistic regression modeling was used to obtain pooled sensitivity and specificity. Meta-regression was performed.

RESULTS

Twenty-two studies with 1182 patients were included. MRI sequences demonstrated high overall diagnostic performance with pooled sensitivity of 91% (95% confidence interval [CI], 87-93%) and specificity of 89% (95% CI, 85-93%). The area under the hierarchical summary receiver operating characteristic curve was 0.92 (95% CI, 0.90-0.94). Studies using DSC or ASL showed high diagnostic performance (sensitivity of 93% [95% CI, 89-97%] and specificity of 91% [95% CI, 86-96%]). Heterogeneity was only detected in specificity (I² = 66.84%) and magnetic field strength was revealed to be a significant factor affecting study heterogeneity.

DATA CONCLUSION

MRI showed overall high diagnostic performance for differentiating PCNSL from glioblastoma, with studies using DSC or ASL showing high diagnostic performance. Therefore, MRI sequences including DSC or ASL is a potential diagnostic tool for differentiating PCNSL from glioblastoma.

LEVEL OF EVIDENCE

3 Technical Efficacy Stage: 2 J. Magn. Reson. Imaging 2019;50:560-572.

Biomarkers Reflecting The Destruction Of The Blood-Brain Barrier Are Valuable In Predicting The Risk Of Lymphomas With Central Nervous System Involvement

OBJECTIVE

We aimed to identify the biomarkers in cerebrospinal fluid (CSF) that facilitate the diagnosis of lymphomas with central nervous system (CNS) involvement.

METHODS

Four cases of non-Hodgkin's lymphoma (NHL) patients with/without CNS involvement were enrolled respectively, and non-CNS tumor patients (n=3) were selected to be the controls. Lab biomarkers, cytokines, and tight junction proteins (TJs) in CSF and serum were measured.

RESULTS

When comparing the CNS to non-CNS group, cytokine including MMP-9 (15.24 vs 0.36 ng/mL), CCL-2 (1922.04 vs 490.68 pg/mL), and sVCAM-1 (61.36 vs 9.00 pg/mL), TJs including OCLN (6.68 vs 2.59 pg/mL), and ZO-1 (710.04 vs 182.98 pg/mL) in CSF were significantly higher in lymphomas patients with CNS involvement than those without CNS involvement. However, serum biomarkers were not significantly elevated. Contrary to the major findings, all conventional biomarkers and MRI results showed no significant change.

CONCLUSION

CSF biomarkers affecting BBB disruption are valuable in mirroring the risk of lymphoma CNS metastasis. Further study with a larger sample size is needed to verify these biomarkers in predicting lymphoma CNS involvement.

Utility of arterial spin labelling MRI for discriminating atypical high-grade glioma from primary central nervous system lymphoma

AIM

To evaluate the ability of arterial spin labelling (ASL) magnetic resonance imaging (MRI) in differentiating primary central nervous system lymphoma (PCNSL) from atypical high-grade glioma (HGG), as well as exploring the underlying pathological mechanisms.

METHODS AND MATERIALS

Twenty-three patients with PCNSL and 17 patients with atypical HGG who underwent ASL-MRI were identified retrospectively. Absolute cerebral blood flow (aCBF) and normalised cerebral blood flow (nCBF) values were obtained, and were compared between PCNSL and atypical HGG using the Mann-Whitney U-test. The performance in discriminating between PCNSL and atypical HGG was evaluated using receiver-operating characteristics analysis and area-under-the-curve (AUC) values for aCBF and nCBF. The correlation between microvessel density (MVD) and aCBF was determined by Spearman's correlation analysis.

RESULTS

Atypical HGG demonstrated significantly higher aCBF, nCBF, and MVD values than PCNSL (p<0.05). The diagnostic accuracy of discriminating PCNSL from atypical HGG showed AUC=0.877 (95% confidence interval [CI] 0.735-0.959) for aCBF, and AUC=0.836 (95% confidence interval [CI] 0.685-0.934) for nCBF. There was a moderate positive correlation between aCBF values of region of interest (ROI >30 mm²) in the enhanced area and MVD values (rho=0.579, p=0.0001), and a strong positive correlation between aCBF values MVD based on "point-to-point biopsy" (rho=0.83, p=0.0029). Interobserver agreements for aCBF and nCBF were excellent (ICC >0.75).

CONCLUSIONS

ASL perfusion MRI is a useful imaging technique for the discrimination between atypical HGG and PCNSL, which may be determined by the difference of MVD between them.

Added Value of Spectroscopy to Perfusion MRI in the Differential Diagnostic Performance of Common Malignant Brain Tumors

BACKGROUND AND PURPOSE

Perfusion and spectroscopic MR imaging provide noninvasive physiologic and metabolic characterization of tissues, which can help in differentiating brain tumors. We investigated the diagnostic role of perfusion and spectroscopic MR imaging using individual and combined classifiers of these modalities and assessed the added performance value that spectroscopy can provide to perfusion using optimal combined classifiers that have the highest differential diagnostic performance to discriminate lymphomas, glioblastomas, and metastases.

MATERIALS AND METHODS

From January 2013 to January 2016, fifty-five consecutive patients with histopathologically proved lymphomas, glioblastomas, and metastases were included after undergoing MR imaging. The perfusion parameters (maximum relative CBV, maximum percentage of signal intensity recovery) and spectroscopic concentration ratios (lactate/Cr, Cho/NAA, Cho/Cr, and lipids/Cr) were analyzed individually and in optimal combinations. Differences among tumor groups, differential diagnostic performance, and differences in discriminatory performance of models with quantification of the added performance value of spectroscopy to perfusion were tested using 1-way ANOVA models, receiver operating characteristic analysis, and comparisons between receiver operating characteristic analysis curves using a bivariate χ², respectively.

RESULTS

The highest differential diagnostic performance was obtained with the following combined classifiers: maximum percentage of signal intensity recovery-Cho/NAA to discriminate lymphomas from glioblastomas and metastases, significantly increasing the sensitivity from 82.1% to 95.7%; relative CBV-Cho/NAA to discriminate glioblastomas from lymphomas and metastases, significantly increasing the specificity from 92.7% to 100%; and maximum percentage of signal intensity recovery-lactate/Cr and maximum percentage of signal intensity recovery-Cho/Cr to discriminate metastases from lymphomas and glioblastomas, significantly increasing the specificity from 83.3% to 97.0% and 100%, respectively.

CONCLUSIONS

Spectroscopy yielded an added performance value to perfusion using optimal combined classifiers of these modalities, significantly increasing the differential diagnostic performances for these common brain tumors.

Differentiating between Glioblastoma and Primary CNS Lymphoma Using Combined Whole-tumor Histogram Analysis of the Normalized Cerebral Blood Volume and the Apparent Diffusion Coefficient

Purpose:

This study aimed to determine whether whole-tumor histogram analysis of normalized cerebral blood volume (nCBV) and apparent diffusion coefficient (ADC) for contrast-enhancing lesions can be used to differentiate between glioblastoma (GBM) and primary central nervous system lymphoma (PCNSL).

Methods:

From 20 patients, 9 with PCNSL and 11 with GBM without any hemorrhagic lesions, underwent MRI, including diffusion-weighted imaging and dynamic susceptibility contrast perfusion-weighted imaging before surgery. Histogram analysis of nCBV and ADC from whole-tumor voxels in contrast-enhancing lesions was performed. An unpaired t-test was used to compare the mean values for each type of tumor. A multivariate logistic regression model (LRM) was performed to classify GBM and PCNSL using the best parameters of ADC and nCBV.

Results:

All nCBV histogram parameters of GBMs were larger than those of PCNSLs, but only average nCBV was statistically significant after Bonferroni correction. Meanwhile, ADC histogram parameters were also larger in GBM compared to those in PCNSL, but these differences were not statistically significant. According to receiver operating characteristic curve analysis, the nCBV average and ADC 25th percentile demonstrated the largest area under the curve with values of 0.869 and 0.838, respectively. The LRM combining these two parameters differentiated between GBM and PCNSL with a higher area under the curve value (Logit (P) = −21.12 + 10.00 × ADC 25th percentile (10⁻³ mm²/s) + 5.420 × nCBV mean, P < 0.001).

Conclusion:

Our results suggest that whole-tumor histogram analysis of nCBV and ADC combined can be a valuable objective diagnostic method for differentiating between GBM and PCNSL.

Machine Learning-based Texture Analysis of Contrast-enhanced MR Imaging to Differentiate between Glioblastoma and Primary Central Nervous System Lymphoma

Purpose:

Although advanced MRI techniques are increasingly available, imaging differentiation between glioblastoma and primary central nervous system lymphoma (PCNSL) is sometimes confusing. We aimed to evaluate the performance of image classification by support vector machine, a method of traditional machine learning, using texture features computed from contrast-enhanced T₁-weighted images.

Methods:

This retrospective study on preoperative brain tumor MRI included 76 consecutives, initially treated patients with glioblastoma (n = 55) or PCNSL (n = 21) from one institution, consisting of independent training group (n = 60: 44 glioblastomas and 16 PCNSLs) and test group (n = 16: 11 glioblastomas and 5 PCNSLs) sequentially separated by time periods. A total set of 67 texture features was computed on routine contrast-enhanced T₁-weighted images of the training group, and the top four most discriminating features were selected as input variables to train support vector machine classifiers. These features were then evaluated on the test group with subsequent image classification.

Results:

The area under the receiver operating characteristic curves on the training data was calculated at 0.99 (95% confidence interval [CI]: 0.96–1.00) for the classifier with a Gaussian kernel and 0.87 (95% CI: 0.77–0.95) for the classifier with a linear kernel. On the test data, both of the classifiers showed prediction accuracy of 75% (12/16) of the test images.

Conclusions:

Although further improvement is needed, our preliminary results suggest that machine learning-based image classification may provide complementary diagnostic information on routine brain MRI.

18F-FDG PET/CT in immunocompetent patients with primary central nervous system lymphoma: Differentiation from glioblastoma and correlation with DWI

OBJECTIVES

¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) is useful for the detection of cancerous lesions, and FDG uptake is related to the apparent diffusion coefficient (ADC) derived from diffusion-weighted imaging (DWI) of extracranial tumors. The purpose of our study was to investigate the ability of FDG PET/CT in distinguishing primary central nervous system lymphoma (PCNSL) from glioblastoma multiforme (GBM) and to explore the relationship between ¹⁸F-FDG uptake and the ADC in patients with PCNSL.

METHODS

We reviewed 92 patients (40 with PCNSL and 52 with GBM) who underwent FDG PET/CT scans at disease onset. The maximum standardized uptake value (SUVmax), tumor to normal contralateral cortex activity (T/N) ratio, SUVmean, metabolic tumor volume (MTV), and total lesion glycolysis (TLG) of tumor lesions were calculated. Receiver operating characteristic (ROC) curves were generated to determine the diagnostic performance for FDG PET-related parameters to differentiate PCNSL from GBM. Twenty-eight patients with PCNSL (with 34 lesions) also underwent diffusion-weighted imaging. Pearson's correlation analysis was used to assess the relation between SUV- and ADC-derived parameters.

RESULTS

The SUVmax, T/N ratio, SUVmean, and TLG values were significantly higher in PCNSL than in GBM. Comparative ROC analysis indicated that the SUVmax had a greater area under the curve (AUC) of 0.910 than the T/N ratio (0.905, P = .85), SUVmean (0.836, P = .0006), or TLG (0.641, P < 0.0001). The T/N ratio had the highest specificity (94.23%) for differentiating PCNSL from GBM, while the SUVmax had the most optimal sensitivity (92.31%). Further combined analysis of the indices did not significantly improve the AUC. Moderate inverse correlations between the SUVmax, SUVmean, TLG, and the ADC ratio (rADC) were found in PCNSLs (r = -0.526, P = .002; r = -0.504, P = .004; and r = -0.483, P = .006; respectively).

CONCLUSIONS

The SUVmax and T/N ratio may be reliable measures for differentiating PCNSLs from GBMs. Additionally, FDG metabolism indices were inversely proportional to the rADCs of PCNSL lesions.

Primary central nervous system lymphoma and atypical glioblastoma: Differentiation using radiomics approach

OBJECTIVES

To evaluate the diagnostic performance of magnetic resonance (MR) radiomics-based machine-learning algorithms in differentiating primary central nervous system lymphoma (PCNSL) from non-necrotic atypical glioblastoma (GBM).

METHODS

Seventy-seven patients (54 individuals with PCNSL and 23 with non-necrotic atypical GBM), diagnosed from January 2009 to April 2017, were enrolled in this retrospective study. A total of 6,366 radiomics features, including shape, volume, first-order, texture, and wavelet-transformed features, were extracted from multi-parametric (post-contrast T1- and T2-weighted, and fluid attenuation inversion recovery images) and multiregional (enhanced and non-enhanced) tumour volumes. These features were subjected to recursive feature elimination and random forest (RF) analysis with nested cross-validation. The diagnostic abilities of a radiomics machine-learning classifier, apparent diffusion coefficient (ADC), and three readers, who independently classified the tumours based on conventional MR sequences, were evaluated using receiver operating characteristic (ROC) analysis. Areas under the ROC curves (AUC) of the radiomics classifier, ADC value, and the radiologists were compared.

RESULTS

The mean AUC of the radiomics classifier was 0.921 (95 % CI 0.825-0.990). The AUCs of the three readers and ADC were 0.707 (95 % CI 0.622-0.793), 0.759 (95 %CI 0.656-0.861), 0.695 (95 % CI 0.590-0.800) and 0.684 (95 % CI0.560-0.809), respectively. The AUC of the radiomics-based classifier was significantly higher than those of the three readers and ADC (p< 0.001 for all).

CONCLUSIONS

Large-scale radiomics with a machine-learning algorithm can be useful for differentiating PCNSL from atypical GBM, and yields a better diagnostic performance than human radiologists and ADC values.

KEY POINTS

• Machine-learning algorithm radiomics can help to differentiate primary central PCNSL from GBM. • This approach yields a higher diagnostic accuracy than visual analysis by radiologists. • Radiomics can strengthen radiologists' diagnostic decisions whenever conventional MRI sequences are available.

Accuracy of Susceptibility-Weighted Imaging and Dynamic Susceptibility Contrast Magnetic Resonance Imaging for Differentiating High-Grade Glioma from Primary Central Nervous System Lymphomas: Meta-Analysis

BACKGROUND AND PURPOSE

It is always a great challenge to distinguish high-grade glioma (HGG) from primary central nervous system lymphoma (PCNSL). We conducted a meta-analysis to assess the performance of dynamic susceptibility contrast (DSC) and susceptibility-weighted imaging (SWI) in differentiating HGGs from PCNSLs.

METHODS

Heterogeneity and threshold effect were evaluated, and sensitivity, specificity (SPE), and areas under the summary receiver operating characteristic curve (SROC) were calculated.

RESULTS

Thirteen studies with a total of 704 participants were included in this meta-analysis. For the dynamic susceptibility contrast (DSC) technique, the pooled sensitivity and SPE were 0.94 and 0.84, respectively. The area under the curve (AUC) under the SROC was 0.9595, and the diagnostic odds ratio was 81.880. For the susceptibility weighted imaging (SWI) technique, the pooled sensitivity and SPE were 0.92 and 0.87, respectively. The AUC under the SROC was 0.9608, and the diagnostic odds ratio was 89.774. A Z test showed that no significant difference in AUC values between DSC and SWI (P = 0.97) was observed. In addition, no obvious public bias was observed in both the DSC (P = 0.69) and SWI (P = 0.32) groups.

CONCLUSIONS

This meta-analysis indicated that both DSC-MRI and SWI have a high and similar level of diagnostic accuracy for differentiating HGGs from PCNSLs.

Diffusion Weighted/Tensor Imaging, Functional MRI and Perfusion Weighted Imaging in Glioblastoma-Foundations and Future

In this article, we review the basics of diffusion tensor imaging and functional MRI, their current utility in preoperative neurosurgical mapping, and their limitations. We also discuss potential future applications, including implementation of resting state functional MRI. We then discuss perfusion and diffusion-weighted imaging and their application in advanced neuro-oncologic practice. We explain how these modalities can be helpful in guiding surgical biopsies and differentiating recurrent tumor from treatment related changes.

Differentiation of glioblastoma multiforme, metastases and primary central nervous system lymphomas using multiparametric perfusion and diffusion MR imaging of a tumor core and a peritumoral zone-Searching for a practical approach

Introduction

In conventional MR examinations glioblastomas multiforme (GBMs), metastases and primary CNS lymphomas (PCNSLs) may show very similar appearance. The aim of the study was to evaluate usefulness of multiparametric T2*DSC perfusion and diffusion MR imaging in the preoperative differentiation of these tumors.

Material and methods

Seventy four solitary enhancing tumors (27 GBMs, 30 metastases, 17 PCNSLs) were enrolled in the study. Parameters of cerebral blood volume (rCBV), peak height (rPH), percentage of signal recovery (rPSR) and apparent diffusion coefficient (ADC) were assessed from the tumor core and the peritumoral non-enhancing T2-hyperintense zone.

Results

Within the tumor core there were no differences in perfusion and diffusion parameters between GBMs and metastases. Compared to GBMs and metastases, PCNSLs showed significantly lower rCBV and rPH, ADC as well as higher rPSR values. Max rCBV with a cut-off value of 2.18 demonstrated the highest accuracy of 0.98 in differentiating PCNSLs from other tumors. To distinguish GBMs from metastases analysis of the peritumoral zone was performed showing significantly higher rCBV, rPH and lower ADC values in GBMs with the highest accuracy of 0.94 found for max rCBV at a cut-off value of 0.98.

Conclusions

Max rCBV seems to be the most important parameter to differentiate GBMs, metastases and PCNSLs. Analysis of max rCBV within the tumor core enables to distinguish hypoperfused PCNSLs from hyperperfused GBMs and metastases while evaluation of max rCBV within the peritumoral zone is helpful to distinguish GBMs showing peritumoral infiltration from metastases surrounded by pure edema.

Leakage correction improves prognosis prediction of dynamic susceptibility contrast perfusion MRI in primary central nervous system lymphoma

To evaluate whether the cerebral blood volume (CBV) measurement with leakage correction from dynamic susceptibility contrast perfusion weighted imaging can be useful in predicting prognosis for primary central nervous system lymphoma (PCNSL). 46 PCNSL patients were included and classified by radiation therapy (RT) stratification into RT (n = 30) and non-RT (n = 16) groups. The corresponding histogram parameters of normalized CBV (nCBV) maps with or without leakage correction were calculated on contrast-enhanced T1 weighted image (CE T1WI) or on fluid attenuated inversion recovery image. The 75^th percentile nCBV with leakage correction based on CE T1WI (T1 nCBVL_75%) had a significant difference between the short and long progression free survival (PFS) subgroups of the RT group and the non-RT group, respectively. Based on the survival analysis, patients in the RT group with high T1 nCBVL_75% had earlier progression than the others with a low T1 nCBVL_75%. However, patients in the non-RT group with a high T1 nCBVL_75% had slower progression than the others with a low T1 nCBVL_75%. Based on RT stratification, the CBV with leakage correction has potential as a noninvasive biomarker for the prognosis prediction of PCNSL to identify high risk patients and it has a different correlation with the PFS based on the presence of combined RT.

Primary central nervous system lymphoma and glioblastoma differentiation based on conventional magnetic resonance imaging by high-throughput SIFT features

PURPOSE OF THE STUDY

Due to the totally different therapeutic regimens needed for primary central nervous system lymphoma (PCNSL) and glioblastoma (GBM), accurate differentiation of the two diseases by noninvasive imaging techniques is important for clinical decision-making.

MATERIALS AND METHODS

Thirty cases of PCNSL and 66 cases of GBM with conventional T1-contrast magnetic resonance imaging (MRI) were analyzed in this study. Convolutional neural networks was used to segment tumor automatically. A modified scale invariant feature transform (SIFT) method was utilized to extract three-dimensional local voxel arrangement information from segmented tumors. Fisher vector was proposed to normalize the dimension of SIFT features. An improved genetic algorithm (GA) was used to extract SIFT features with PCNSL and GBM discrimination ability. The data-set was divided into a cross-validation cohort and an independent validation cohort by the ratio of 2:1. Support vector machine with the leave-one-out cross-validation based on 20 cases of PCNSL and 44 cases of GBM was employed to build and validate the differentiation model.

RESULTS

Among 16,384 high-throughput features, 1356 features show significant differences between PCNSL and GBM with p < 0.05 and 420 features with p < 0.001. A total of 496 features were finally chosen by improved GA algorithm. The proposed method produces PCNSL vs. GBM differentiation with an area under the curve (AUC) curve of 99.1% (98.2%), accuracy 95.3% (90.6%), sensitivity 85.0% (80.0%) and specificity 100% (95.5%) on the cross-validation cohort (and independent validation cohort).

CONCLUSIONS

Since the local voxel arrangement characterization provided by SIFT features, proposed method produced more competitive PCNSL and GBM differentiation performance by using conventional MRI than methods based on advanced MRI.