Search

Purpose: This study aimed to compare assessments by radiologists, artificial intelligence (AI), and quantitative measurement using synthetic MRI (SyMRI) for differential diagnosis between astrocytoma, IDH-mutant and oligodendroglioma, and IDH-mutant and 1p/19q-codeleted and to identify the superior method., Methods: Thirty-three cases (men, 14; women, 19) comprising 19 astrocytomas and 14 oligodendrogliomas were evaluated. Four radiologists independently evaluated the presence of the T2-FLAIR mismatch sign. A 3D convolutional neural network (CNN) model was trained using 50 patients outside the test group (28 astrocytomas and 22 oligodendrogliomas) and transferred to evaluate the T2-FLAIR mismatch lesions in the test group. If the CNN labeled more than 50% of the T2-prolonged lesion area, the result was considered positive. The T1/T2-relaxation times and proton density (PD) derived from SyMRI were measured in both gliomas. Each quantitative parameter (T1, T2, and PD) was compared between gliomas using the Mann-Whitney U-test. Receiver-operating characteristic analysis was used to evaluate the diagnostic performance., Results: The mean sensitivity, specificity, and area under the curve (AUC) of radiologists vs. AI were 76.3% vs. 94.7%; 100% vs. 92.9%; and 0.880 vs. 0.938, respectively. The two types of diffuse gliomas could be differentiated using a cutoff value of 2290/128 ms for a combined 90 th percentile of T1 and 10 th percentile of T2 relaxation times with 94.4/100% sensitivity/specificity with an AUC of 0.981., Conclusion: Compared to the radiologists' assessment using the T2-FLAIR mismatch sign, the AI and the SyMRI assessments increased both sensitivity and objectivity, resulting in improved diagnostic performance in differentiating gliomas., (© 2024. The Author(s).)

Objectives: To develop an automated model to detect brain metastases using a convolutional neural network (CNN) and volume isotropic simultaneous interleaved bright-blood and black-blood examination (VISIBLE) and to compare its diagnostic performance with the observer test., Methods: This retrospective study included patients with clinical suspicion of brain metastases imaged with VISIBLE from March 2016 to July 2019 to create a model. Images with and without blood vessel suppression were used for training an existing CNN (DeepMedic). Diagnostic performance was evaluated using sensitivity and false-positive results per case (FPs/case). We compared the diagnostic performance of the CNN model with that of the twelve radiologists., Results: Fifty patients (30 males and 20 females; age range 29-86 years; mean 63.3 ± 12.8 years; a total of 165 metastases) who were clinically diagnosed with brain metastasis on follow-up were used for the training. The sensitivity of our model was 91.7%, which was higher than that of the observer test (mean ± standard deviation; 88.7 ± 3.7%). The number of FPs/case in our model was 1.5, which was greater than that by the observer test (0.17 ± 0.09)., Conclusions: Compared to radiologists, our model created by VISIBLE and CNN to diagnose brain metastases showed higher sensitivity. The number of FPs/case by our model was greater than that by the observer test of radiologists; however, it was less than that in most of the previous studies with deep learning., Key Points: • Our convolutional neural network based on bright-blood and black-blood examination to diagnose brain metastases showed a higher sensitivity than that by the observer test. • The number of false-positives/case by our model was greater than that by the previous observer test; however, it was less than those from most previous studies. • In our model, false-positives were found in the vessels, choroid plexus, and image noise or unknown causes., (© 2021. The Author(s), under exclusive licence to European Society of Radiology.)

Purpose: Diffusion-weighted imaging (DWI) plays an important role in the preoperative assessment of gliomas; however, the diagnostic performance of histogram-derived parameters from mono-, bi-, and stretched-exponential DWI models in the grading of gliomas has not been fully investigated. Therefore, we compared these models' ability to differentiate between high-grade and low-grade gliomas., Methods: This retrospective study included 22 patients with diffuse gliomas (age, 23-74 years; 12 males; 11 high-grade and 11 low-grade gliomas) who underwent preoperative 3 T-magnetic resonance imaging from October 2014 to August 2019. The apparent diffusion coefficient was calculated from the mono-exponential model. Using 13 b-values, the true-diffusion coefficient, pseudo-diffusion coefficient, and perfusion fraction were obtained from the bi-exponential model, and the distributed-diffusion coefficient and heterogeneity index were obtained from the stretched-exponential model. Region-of-interests were drawn on each imaging parameter map for subsequent histogram analyses., Results: The skewness of the apparent diffusion, true-diffusion, and distributed-diffusion coefficients was significantly higher in high-grade than in low-grade gliomas (0.67 ± 0.67 vs. - 0.18 ± 0.63, 0.68 ± 0.74 vs. - 0.08 ± 0.66, 0.63 ± 0.72 vs. - 0.15 ± 0.73; P = 0.0066, 0.0192, and 0.0128, respectively). The 10th percentile of the heterogeneity index was significantly lower (0.77 ± 0.08 vs. 0.88 ± 0.04; P = 0.0004), and the 90th percentile of the perfusion fraction was significantly higher (12.64 ± 3.44 vs. 7.14 ± 1.70%: P < 0.0001), in high-grade than in low-grade gliomas. The combination of the 10th percentile of the true-diffusion coefficient and 90th percentile of the perfusion fraction showed the best area under the receiver operating characteristic curve (0.96)., Conclusion: The bi-exponential model exhibited the best diagnostic performance for differentiating high-grade from low-grade gliomas.

Aim: To evaluate the diagnostic feasibility of probabilistic analysis using voxel-based morphometry (VBM) in differentiating primary central nervous system lymphoma (PCNSL) from glioblastoma (GBM)., Materials and Methods: In total, 118 patients with GBM (57 males, 61 females; mean [± standard deviation] age, 56.9±19.3 years; median, 61 years) and 52 patients with PCNSL (37 males, 15 females; mean age, 62±13.3 years, median, 66 years) were studied retrospectively. Each patient underwent preoperative contrast-enhanced T1-weighted imaging (CE-T1WI) using a 1.5 or 3 T magnetic resonance imaging (MRI) system. To assess preferential occurrence sites, images from CE-T1WI were co-registered and spatially normalised using the MNI152 T1 template. Subsequently, a region of interest (ROI) was placed in the centre of the enhancing tumour in normalised images with 1-mm isotropic resolution. The same ROI between normalised and T1 template images was set up using an ROI manager function in ImageJ software. A spherical volume of interest (VOI) with a radius of 10 mm was determined. A probability map was created by overlaying each image with the VOI. Each VOI was removed from T1 template images for VBM analysis. VBM analysis was performed using statistical parametric mapping (SPM) 12 software under default settings., Results: VBM analysis showed significantly higher frequency in the splenium of the corpus callosum among PCNSL patients than among GBM patients (p<0.05; family-wise error correction)., Conclusion: Topographic analysis using VBM provides useful information for differentiating PCNSL from GBM., (Copyright © 2019 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.)

Purpose: The purpose of this study was to identify magnetic resonance imaging (MRI) features that are associated with telomerase reverse transcriptase promoter mutation (TERTm) in glioblastoma., Materials and Methods: A total of 112 patients with glioblastoma who had MRI at 1.5- or 3.0-T were retrospectively included. There were 43 patients with glioblastoma with wild-type TERT (TERTw) (22 men, 21 women; mean age, 47±25 [SD] years; age range: 3-84 years) and 69 patients with glioblastoma with TERTm (34 men, 35 women; mean age 64±11 [SD] years; age range, 41--85 years). The feature vectors consist of 11 input units for two clinical parameters (age and gender) and nine MRI characteristics (tumor location, subventricular extension, cortical extension, multiplicity, enhancing volume, necrosis volume, the percentage of necrosis volume, minimum apparent diffusion coefficient [ADC] and normalized ADC). First, the diagnostic performance using univariate and multivariate logistic regression analyses was evaluated. Second, the cross-validation of the support vector machine (SVM) was performed by using leave-one-out method with 43 TERTw and 69 TERTm to evaluate the diagnostic performance. In addition, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy for the differentiation between TERTw and TERTm were compared between logistic regression analysis and SVM., Results: With multivariate analysis, the percentage of necrosis volume and age were significantly greater in TERTm glioblastoma than in TERTw glioblastoma. SVM allowed discriminating between TERTw glioblastoma and TERTm glioblastoma with sensitivity, specificity, PPV, NPV, and accuracy of 85.7% [60/70; 95% confidence interval (CI): 75.3-92.9%], 54.8% (23/42; 95% CI: 38.7-70.2%), 75.9% (60/79; 95% CI: 69.1-81.7%), 69.7% (23/33; 95% CI: 54.9-81.3%) and 74.1% (83/112; 95% CI: 65.0-81.9%), respectively., Conclusion: The percentage of necrosis volume and age may surrogate for predicting TERT mutation status in glioblastoma., (Copyright © 2019 Société française de radiologie. Published by Elsevier Masson SAS. All rights reserved.)

Background and Purpose: Intravoxel incoherent motion imaging, which simultaneously measures diffusion and perfusion parameters, is promising for brain tumor grading. However, intravoxel incoherent motion imaging has not been tested in children. The purpose of this study was to evaluate the correlation between intravoxel incoherent motion parameters and histology to assess the accuracy of intravoxel incoherent motion imaging for pediatric intracranial tumor grading., Materials and Methods: Between April 2013 and September 2015, 17 children (11 boys, 6 girls; 2 months to 15 years of age) with intracranial tumors were included in this retrospective study. Intravoxel incoherent motion parameters were fitted using 13 b-values for a biexponential model. The perfusion-free diffusion coefficient, pseudodiffusion coefficient, and perfusion fraction were measured in high- and low-grade tumors. These intravoxel incoherent motion parameters and the ADC were compared using the unpaired t test. The correlations between the intravoxel incoherent motion parameters and microvessel density or the MIB-1 index were analyzed using the Spearman correlation test. Receiver operating characteristic analysis was used to evaluate diagnostic performance., Results: The perfusion-free diffusion coefficient and ADC were lower in high-grade than in low-grade tumors (perfusion-free diffusion coefficient, 0.85 ± 0.40 versus 1.53 ± 0.21 × 10 -3 mm 2 /s, P < .001; ADC, 1.04 ± 0.33 versus 1.60 ± 0.21 × 10 -3 mm 2 /s, P < .001). The pseudodiffusion coefficient showed no difference between the groups. The perfusion fraction was higher in high-grade than in low-grade tumors (21.7 ± 8.2% versus 7.6 ± 4.3%, P < .001). Receiver operating characteristic analysis found that the combined perfusion-free diffusion coefficient and perfusion fraction had the best diagnostic performance for tumor differentiation (area under the curve = 0.986)., Conclusions: Intravoxel incoherent motion imaging reflects tumor histology and may be a helpful, noninvasive method for pediatric intracranial tumor grading., (© 2019 by American Journal of Neuroradiology.)

Objective: To evaluate the quantification performance of the perfusion fraction (f) measured with intravoxel incoherent motion (IVIM) MR imaging in a comparison with the histological vascular density in meningiomas., Methods: 29 consecutive patients with meningioma (59.0 ± 16.8 years old, 8 males and 21 females) who underwent a subsequent surgical resection were examined with both IVIM imaging and a histopathological analysis. IVIM imaging was conducted using a single-shot SE-EPI sequence with 13 b-factors (0, 10, 20, 30, 50, 80, 100, 200, 300, 400, 600, 800, 1000 s mm - 2 ) at 3T. The perfusion fraction (f) was calculated by fitting the IVIM bi-exponential model. The 90-percentile f-value in the tumor region-of-interest (ROI) was defined as the maximum f-value (f-max). Histopathological vascular density (%Vessel) was measured on CD31-immunostainted histopathological specimens. The correlation and agreement between the f-values and %Vessel was assessed., Results: The f-max (15.5 ± 5.5%) showed excellent agreement [intraclass correlation coefficient (ICC) = 0.754] and a significant correlation (r = 0.69, p < 0.0001) with the %Vessel (12.9 ± 9.4%) of the tumors. The Bland-Altman plot analysis showed excellent agreement between the f-max and %Vessel (bias, -2.6%; 95% limits of agreement, from -16.0 to 10.8%). The f-max was not significantly different among the histological subtypes of meningioma., Conclusion: An excellent agreement and a significant correlation were observed between the f-values and %Vessel. The f-value can be used as a noninvasive quantitative imaging measure to directly assess the vascular volume fraction in brain tumors. Advances in knowledge: The f-value measured by IVIM imaging showed a significant correlation and an excellent agreement with the histological vascular density in the meningiomas. The f-value can be used as a noninvasive and quantitative imaging measure to directly assess the volume fraction of capillaries in brain tumors.