Search

Identification of genotypes is crucial for treatment of glioma. Here, we developed a method to predict tumor genotypes using a pretrained convolutional neural network (CNN) from magnetic resonance (MR) images and compared the accuracy to that of a diagnosis based on conventional radiomic features and patient age. Multisite preoperative MR images of 164 patients with grade II/III glioma were grouped by IDH and TERT promoter (pTERT) mutations as follows: (1) IDH wild type, (2) IDH and pTERT co-mutations, (3) IDH mutant and pTERT wild type. We applied a CNN (AlexNet) to four types of MR sequence and obtained the CNN texture features to classify the groups with a linear support vector machine. The classification was also performed using conventional radiomic features and/or patient age. Using all features, we succeeded in classifying patients with an accuracy of 63.1%, which was significantly higher than the accuracy obtained from using either the radiomic features or patient age alone. In particular, prediction of the pTERT mutation was significantly improved by the CNN texture features. In conclusion, the pretrained CNN texture features capture the information of IDH and TERT genotypes in grade II/III gliomas better than the conventional radiomic features.

Molecular biological characterization of tumors has become a pivotal procedure for glioma patient care. The aim of this study is to build conventional MRI-based radiomics model to predict genetic alterations within grade II/III gliomas attempting to implement lesion location information in the model to improve diagnostic accuracy. One-hundred and ninety-nine grade II/III gliomas patients were enrolled. Three molecular subtypes were identified: IDH1/2-mutant, IDH1/2-mutant with TERT promoter mutation, and IDH-wild type. A total of 109 radiomics features from 169 MRI datasets and location information from 199 datasets were extracted. Prediction modeling for genetic alteration was trained via LASSO regression for 111 datasets and validated by the remaining 58 datasets. IDH mutation was detected with an accuracy of 0.82 for the training set and 0.83 for the validation set without lesion location information. Diagnostic accuracy improved to 0.85 for the training set and 0.87 for the validation set when lesion location information was implemented. Diagnostic accuracy for predicting 3 molecular subtypes of grade II/III gliomas was 0.74 for the training set and 0.56 for the validation set with lesion location information implemented. Conventional MRI-based radiomics is one of the most promising strategies that may lead to a non-invasive diagnostic technique for molecular characterization of grade II/III gliomas.

Purpose: Arterial spin labeling (ASL) is a non-invasive perfusion technique that may be an alternative to dynamic susceptibility contrast magnetic resonance imaging (DSC-MRI) for assessment of brain tumors. To our knowledge, there have been no reports on histogram analysis of ASL. The purpose of this study was to determine whether ASL is comparable with DSC-MRI in terms of differentiating high-grade and low-grade gliomas by evaluating the histogram analysis of cerebral blood flow (CBF) in the entire tumor., Methods: Thirty-four patients with pathologically proven glioma underwent ASL and DSC-MRI. High-signal areas on contrast-enhanced T 1 -weighted images or high-intensity areas on fluid-attenuated inversion recovery images were designated as the volumes of interest (VOIs). ASL-CBF, DSC-CBF, and DSC-cerebral blood volume maps were constructed and co-registered to the VOI. Perfusion histogram analyses of the whole VOI and statistical analyses were performed to compare the ASL and DSC images., Results: There was no significant difference in the mean values for any of the histogram metrics in both of the low-grade gliomas (n = 15) and the high-grade gliomas (n = 19). Strong correlations were seen in the 75th percentile, mean, median, and standard deviation values between the ASL and DSC images. The area under the curve values tended to be greater for the DSC images than for the ASL images., Conclusions: DSC-MRI is superior to ASL for distinguishing high-grade from low-grade glioma. ASL could be an alternative evaluation method when DSC-MRI cannot be used, e.g., in patients with renal failure, those in whom repeated examination is required, and in children.

Reports have suggested that tumor textures presented on T2-weighted images correlate with the genetic status of glioma. Therefore, development of an image analyzing framework that is capable of objective and high throughput image texture analysis for large scale image data collection is needed. The current study aimed to address the development of such a framework by introducing two novel parameters for image textures on T2-weighted images, i.e., Shannon entropy and Prewitt filtering. Twenty-two WHO grade 2 and 28 grade 3 glioma patients were collected whose pre-surgical MRI and IDH1 mutation status were available. Heterogeneous lesions showed statistically higher Shannon entropy than homogenous lesions (p = 0.006) and ROC curve analysis proved that Shannon entropy on T2WI was a reliable indicator for discrimination of homogenous and heterogeneous lesions (p = 0.015, AUC = 0.73). Lesions with well-defined borders exhibited statistically higher Edge mean and Edge median values using Prewitt filtering than those with vague lesion borders (p = 0.0003 and p = 0.0005 respectively). ROC curve analysis also proved that both Edge mean and median values were promising indicators for discrimination of lesions with vague and well defined borders and both Edge mean and median values performed in a comparable manner (p = 0.0002, AUC = 0.81 and p < 0.0001, AUC = 0.83, respectively). Finally, IDH1 wild type gliomas showed statistically lower Shannon entropy on T2WI than IDH1 mutated gliomas (p = 0.007) but no difference was observed between IDH1 wild type and mutated gliomas in Edge median values using Prewitt filtering. The current study introduced two image metrics that reflect lesion texture described on T2WI. These two metrics were validated by readings of a neuro-radiologist who was blinded to the results. This observation will facilitate further use of this technique in future large scale image analysis of glioma., Competing Interests: The authors have declared that no competing interests exist.

Object: Immunotherapy targeting the Wilms tumor 1 (WT1) gene product is a promising treatment modality for patients with malignant gliomas, and there have been reports of encouraging results. It has become clear, however, that Gd-enhanced MR imaging does not reflect prognosis, thereby necessitating a more robust imaging evaluation system for monitoring response to WT1 immunotherapy. To meet this demand, the authors performed a voxel-wise parametric response map (PRM) analysis of (11)C-methionine PET (MET-PET) in WT1 immunotherapy and compared the data with the overall survival after initiation of WT1 immunotherapy (OS(WT1))., Methods: Fourteen patients with recurrent malignant glioma were included in the study, and OS(WT1) was compared with: 1) volume and length change in the contrast area of the tumor on Gd-enhanced MR images; 2) change in maximum uptake of (11)C-methionine; and 3) a more detailed voxel-wise PRM analysis of MET-PET pre- and post-WT1 immunotherapy., Results: The PRM analysis was able to identify the following 3 areas within the tumor core: 1) area with no change in (11)C-methionine uptake pre- and posttreatment; 2) area with increased (11)C-methionine uptake posttreatment (PRM(+MET)); and 3) area with decreased (11)C-methionine uptake posttreatment. While the results of Gd-enhanced MR imaging volumetric and conventional MET-PET analysis did not correlate with OS(WT1) (p = 0.270 for Gd-enhanced MR imaging length, p = 0.960 for Gd-enhanced MR imaging volume, and p = 0.110 for MET-PET), the percentage of PRM(+MET) area showed excellent correlation (p = 0.008) with OS(WT1)., Conclusions: This study describes the limited value of Gd-enhanced MR imaging and highlights the potential of voxel-wise PRM analysis of MET-PET for monitoring treatment response in immunotherapy for malignant gliomas. Clinical trial registration no.: UMIN000002001.

Brain Tumor Image Analysis (BraTumIA) is a fully automated segmentation tool dedicated to detecting brain tumors imaged by magnetic resonance imaging (MRI). BraTumIA has recently been applied to several clinical investigations; however, the validity of this novel method has not yet been fully examined. The present study was conducted to validate the quality of tumor segmentation with BraTumIA in comparison with results from 11C-methionine positron emission tomography (MET-PET). A total of 45 consecutive newly diagnosed high-grade gliomas imaged by MRI and MET-PET were analyzed. Automatic tumor segmentation was conducted by BraTumIA and the resulting segmentation images were registered to MET-PET. Three-dimensional conformal association between these two modalities was calculated, considering MET-PET as the gold standard. High underestimation and overestimation errors were observed in tumor segmentation calculated by BraTumIA compared with MET-PET. Furthermore, when the tumor/normal ratio threshold was set at 1.3 from MET-PET, the BraTumIA false-positive fraction was ~0.4 and the false-negative fraction was 0.9. By tightening this threshold to 2.0, the BraTumIA false-positive fraction was 0.6 and the false-negative fraction was 0.6. Following comparison of segmentation performance with BraTumIA with regard to glioblastoma (GBM) and World Health Organization (WHO) grade III glioma, GBM exhibited better segmentation compared with WHO grade III glioma. Although BraTumIA may be able to detect enhanced tumors, non-enhancing tumors and necrosis, the spatial concordance rate with MET-PET was relatively low. Careful interpretation is therefore required when using this technique. [ABSTRACT FROM AUTHOR]

A number of studies have revealed the usefulness of multimodal imaging in gliomas. Although the results have been heavily affected by the method used for region of interest (ROI) design, the most discriminatory method for setting the ROI remains unclear. The aim of the present study was to determine the most suitable ROI design for 18F-fluorodeoxyglucose (FDG) and 11C-methionine (MET) positron emission tomography (PET), apparent diffusion coefficient (ADC), and fractional anisotropy (FA) obtained by diffusion tensor imaging (DTI) from the viewpoint of grades of non-enhancing gliomas. A total of 31 consecutive patients with newly diagnosed, histologically confirmed magnetic resonance (MR) non-enhancing gliomas who underwent FDG-PET, MET-PET and DTI were retrospectively investigated. Quantitative measurements were performed using four different ROIs; hotspot/tumor center and whole tumor, constructed in either two-dimensional (2D) or three-dimensional (3D). Histopathological grading of the tumor was considered as empirical truth and the quantitative measurements obtained from each ROI was correlated with the grade of the tumor. The most discriminating ROI for non-enhancing glioma grading was different according to the different imaging modalities. 2D-hotspot/center ROI was most discriminating for FDG-PET (P=0.087), ADC map (P=0.0083), and FA map (P=0.25), whereas 3D-whole tumor ROI was best for MET-PET (P=0.0050). In the majority of scenarios, 2D-ROIs performed better than 3D-ROIs. Results from the image analysis using FDG-PET, MET-PET, ADC and FA may be affected by ROI design and the most discriminating ROI for non-enhancing glioma grading was different according to the imaging modality. [ABSTRACT FROM AUTHOR]

Purpose: This study aimed to compare the tumor volume definition using conventional magnetic resonance (MR) and 11C-methionine positron emission tomography (MET/PET) images in the differentiation of the pre-operative glioma grade by using whole-tumor histogram analysis of normalized cerebral blood volume (nCBV) maps.Materials and Methods: Thirty-four patients with histopathologically proven primary brain low-grade gliomas (n = 15) and high-grade gliomas (n = 19) underwent pre-operative or pre-biopsy MET/PET, fluid-attenuated inversion recovery, dynamic susceptibility contrast perfusion-weighted magnetic resonance imaging, and contrast-enhanced T1-weighted at 3.0 T. The histogram distribution derived from the nCBV maps was obtained by co-registering the whole tumor volume delineated on conventional MR or MET/PET images, and eight histogram parameters were assessed.Results: The mean nCBV value had the highest AUC value (0.906) based on MET/PET images. Diagnostic accuracy significantly improved when the tumor volume was measured from MET/PET images compared with conventional MR images for the parameters of mean, 50th, and 75th percentile nCBV value (p = 0.0246, 0.0223, and 0.0150, respectively).Conclusion: Whole-tumor histogram analysis of CBV map provides more valuable histogram parameters and increases diagnostic accuracy in the differentiation of pre-operative cerebral gliomas when the tumor volume is derived from MET/PET images. [ABSTRACT FROM AUTHOR]

Background: Ischemic stroke is one form of cancer-associated thrombosis that can greatly worsen a patient's performance status. The present investigation aimed to elucidate the characteristic distribution pattern(s) of cryptogenic stroke lesions using a voxel-based lesion-mapping technique and examine the differences in clinical manifestations between cryptogenic and conventional strokes in patients with advanced cancer.Methods: Data from 43 patients with advanced cancer who developed acute ischemic stroke were retrospectively collected. Stroke etiology was grouped into either cryptogenic or conventional stroke etiology according to the ASCO stroke score. Clinical data were reviewed, and voxel-based lesion mapping using diffusion-weighted imaging (DWI) was performed to visualize the cross-patient spatial distribution of the lesions.Results: Of the 43 patients, 25 were classified as having cryptogenic stroke etiology and 18 were classified as having conventional stroke etiology. Median survival time of patients from stroke onset was 96 days for cryptogenic stroke etiology and 570 days for conventional stroke etiology (P = .01). D-dimer of patients was significantly higher in cryptogenic stoke etiology than in conventional stroke etiology (P = .006). Voxel-based lesion mapping showed that DWI hyperintense lesions accumulated at cortical and internal watershed areas of the cerebrum and at the vascular border zone of the superior cerebellar and posterior inferior cerebellar arteries at the cerebellum.Conclusions: Voxel-based lesion mapping for cryptogenic stroke in patients with advanced cancer showed that lesions accumulated at vascular border zones within the brain both at the cerebrum and at the cerebellum, but not at perforating arterial territories. [ABSTRACT FROM AUTHOR]