Postcontrast T1-weighted Images Research Articles

MRI-based radiomics signature and clinical factor for predicting H3K27M mutation in pediatric high-grade gliomas located in the midline of the brain.

To develop a nomogram based on MRI radiomics and clinical features for preoperatively predicting H3K27M mutation in pediatric high-grade gliomas (pHGGs) with a midline location of the brain. The institutional database was reviewed to identify patients with pHGGs with a midline location of the brain who underwent tumor biopsy with preoperative MRI scans between June 2016 and June 2021. A total of 107 patients with pHGGs, including 79 patients with H3K27M mutation, were consecutively included and randomly divided into training and test sets. Radiomics features were extracted from fluid-attenuated inversion recovery (FLAIR), diffusion-weighted (DW) and post-contrast T1-weighted images, and apparent diffusion coefficient (ADC) maps. The minimum redundancy maximum relevance (MRMR) and least absolute shrinkage and selection operator (LASSO) logistic regression were performed for radiomics signature construction. Clinical and radiological features were analyzed to select clinical predictors. A nomogram was then developed by incorporating the radiomics signature and selected clinical predictors. Nine radiomics features were selected to construct the radiomics signature, which showed a favorable discriminatory ability in training and test sets with an area under the curve (AUC) of 0.95 and 0.92, respectively. Ring enhancement was identified as an independent clinical predictor (p < 0.01). The nomogram, constructed with radiomics signature and ring enhancement, showed good calibration and discrimination in training and testing sets (AUC: 0.95 and 0.90 respectively). The nomogram which combined radiomics signature and ring enhancement had a satisfactory ability to predict H3K27M mutation in pHGGs with a midline of the brain. • Conventional MRI features were not powerful enough to predict H3K27M mutation status in pediatric high-grade gliomas (pHGGs) with a midline location of the brain. • An MRI-based radiomics signature showed satisfactory ability to predict H3K27M mutation status of pHGGs located in the midline of the brain. • Associating the radiomics signature with clinical factors improved predictive performance.

Comparison of Grayscale Value in T1-Weighted Pre- and Post-Contrast Brain MRI Images: with and without Fat Suppression Technique

The MRI T1-weighted image can provide information on the pre- and post-contrast images. Post-contrast images is an image obtained after the administration of GBCA In some cases, not all post-contrast images can show clear lesions so it requires additional technique in the form of Fat Suppression (FS), which works by suppressing the fat signal in an image. The T1-weighted images with and without FS have a different signal intensity. Therefore, the purpose of this study is to compare the signal intensity of the pre- and post-contrast T1-weighted images with and without the FS technique. The signal intensities are indicated with a grayscale value. There are seven T1-weighted images with FS and seven T1-weighted images without FS. Each of the image have a pre-and post-contrast. Image reading is done by a radiology specialist. Area plot was performed on abnormal tissues in each image. Each area will be measured with an ImageJ software to obtain the grayscale mean value. The measurements of the post contrast T1-weighted image showed an increase in the grayscale mean value with or without the FS technique. This showed that the administration of GBCA can increase the signal intensity on the T1-weighted images with or without the FS technique.

Construction of an MRI-based decision tree to differentiate autoimmune and autoinflammatory inner ear disease from chronic otitis media with sensorineural hearing loss

Autoimmune and autoinflammatory inner ear diseases (AIED/AID) are characterized by the symptom of sensorineural hearing loss (SNHL). To date, standardized diagnostic tools for AIED/AID are lacking, and clinically differentiating AIED/AID from chronic otitis media (COM) with SNHL is challenging. This retrospective study aimed to construct a magnetic resonance imaging (MRI)-based decision tree using classification and regression tree (CART) analysis to distinguish AIED/AID from COM. In total, 67 patients were enrolled between January 2004 and October 2019, comprising AIED/AID (n = 18), COM (n = 24), and control groups (n = 25). All patients underwent 3 T temporal bone MRI, including post-contrast T1-weighted images (postT1WI) and post-contrast FLAIR images (postFLAIR). Two radiologists evaluated the presence of otomastoid effusion and inner ear contrast-enhancement on MRI. A CART decision tree model was constructed using MRI features to differentiate AIED/AID from COM and control groups, and diagnostic performance was analyzed. High-intensity bilateral effusion (61.1%) and inner ear enhancement (postFLAIR, 93.8%; postT1WI, 61.1%) were the most common findings in the AIED/AID group. We constructed two CART decision tree models; the first used effusion amount as the first partitioning node and postT1WI-inner ear enhancement as the second node, whereas the second comprised two partitioning nodes with the degree of postFLAIR-enhancement of the inner ear. The first and second models enabled distinction of AIED/AID from COM with high specificity (100% and 94.3%, respectively). The amount of effusion and the degree of inner ear enhancement on MRI may facilitate the distinction between AIED/AID and COM with SNHL using decision tree models, thereby contributing to early diagnosis and intervention.

Primary central nervous system lymphoma: advances in MRI and PET imaging.

Contrast enhanced magnetic resonance imaging (CE-MRI) remains the imaging modality of choice for initial workup, staging, and response assessment after therapy in patients with primary central nervous system lymphoma (PCNSL). While CE-MRI is a sensitive test to detect blood brain barrier (BBB) dysfunction, it does not biologically represent the true tumor burden. Current response assessment criteria relies heavily on two dimensional anatomical measurements on post contrast T1-weighted (T1W) images, as well as pre-contrast T2-weighted (T2W) imaging. Additional MRI features, such as diffusion-weighted imaging (DWI) and perfusion weighted imaging, can be routinely obtained at most centers with MRI capabilities. Emerging evidence supports the incorporation of these data to better define tumor physiology and provide additional valuable clinical tools capable of identifying high risk subgroups as well as early predictors of response to therapies. Further, novel advanced molecular and pathophysiologic characterization of PCNSL provides insights into promising targeted therapeutic approaches. However, significant institutional imaging variation and inconsistent clinical trial reporting diminishes the reliability, reproducibility and eventual translation in day to day management of patients with PCNSL. Here we review established neuroimaging concepts and provide an overview of published literature about novel imaging techniques that may improve diagnosis and response assessments. Finally, we highlight the need for standardization of image acquisition, post-processing, and incorporation of novel imaging biomarkers in early phase PCNSL clinical trials.

MR-Based Radiomics for Differential Diagnosis between Cystic Pituitary Adenoma and Rathke Cleft Cyst.

Background It is often tricky to differentiate cystic pituitary adenoma from Rathke cleft cyst with visual inspection because of similar MRI presentations between them. We aimed to design an MR-based radiomics model for improving differential diagnosis between them. Methods Conventional diagnostic MRI data (T1-,T2-, and postcontrast T1-weighted MR images) were obtained from 215 pathologically confirmed patients (105 cases with cystic pituitary adenoma and the other 110 cases with Rathke cleft cyst) and were divided into training (n = 172) and test sets (n = 43). MRI radiomics features were extracted from the imaging data, and semantic imaging features (n = 15) were visually estimated by two radiologists. Four classifiers were used to construct radiomics models through 5-fold crossvalidation after feature selection with least absolute shrinkage and selection operator. An integrated model by combining radiomics and semantic features was further constructed. The diagnostic performance was validated in the test set. Receiver operating characteristic curve was used to evaluate and compare the performance of the models at the background of diagnostic performance by radiologist. Results In test set, the combined radiomics and semantic model using ANN classifier obtained the best classification performance with an AUC of 0.848 (95% CI: 0.750-0.946), accuracy of 76.7% (95% CI: 64.1-89.4%), sensitivity of 73.9% (95% CI: 56.0-91.9%), and specificity of 80.0% (95% CI: 62.5-97.5%) and performed better than multiparametric model (AUC = 0.792, 95% CI: 0.674-0.910) or semantic model (AUC = 0.823, 95% CI: 0.705-0.941). The two radiologists had an accuracy of 69.8% and 74.4%, respectively, sensitivity of 69.6% and 73.9%, and specificity of 70.0% and 75.0%. Conclusions The MR-based radiomics model had technical feasibility and good diagnostic performance in the differential diagnosis between cystic pituitary adenoma and Rathke cleft cyst.

Role of contrast-enhanced FLAIR MRI in diagnosis of intracranial lesions

BackgroundMR imaging plays a significant role in detection and characterization of different brain diseases. The role of the post-contrast T1-weighted image magnetic resonance imaging (T1W MRI) sequence has been widely established in previous studies and clinical practice. In this study, we aim to share our experience as regards the added value of contrast-enhanced fluid-attenuated inversion recovery (CE-FLAIR) sequence in the diagnosis of various intracranial pathological conditions and evaluate its usefulness in comparison with post-contrast T1W images.ResultsBased on the final radiological diagnosis, the total cases were subdivided into three categories, and the majority of our cases were tumors (81.2%), followed by multiple sclerosis (11.8%), and the least was central nervous system infection (7.1%). CE-FLAIR showed superior enhancement in 35 cases (50.7) and equal enhancement in 25 cases (36.3%). However, it showed less enhancement than post-contrast T1W images in 9 cases (13%). Excellent inter-observer agreement (97.65%) was noted. Regarding lesion conspicuity, good delineation was found in the majority of cases (64.7%), fair delineation in 12.9%, and no delineation in 22.4%. A statistically significant difference was found in signal intensity of lesion between pre- and post-contrast FLAIR sequences. Contrast to background ratio was statistically significant in CE FLAIR images in comparison to CE T1 images.ConclusionCE-FLAIR imaging should be used as a routine or adjunctive sequence to CE-T1WI to enhance early detection and increase the diagnostic confidence in MRI examination of different brain pathological conditions.

Early stage glioblastoma: retrospective multicentric analysis of clinical and radiological features.

The aim of this study was to report our experience with early stage glioblastoma (e-GB) and to investigate the possible clinical and imaging features that may be helpful to the radiologist to correctly diagnose this entity. We performed a retrospective research of patients diagnosed with glioblastoma at two hospitals during a 10-year period. We reviewed all pre-operative MR and included only patients with early stage GB lesions, characterized by hyperintense on T2-weighted signal, with or without contrast-enhancement at post-contrast T1-weighted images, without "classic" imaging appearance of GB (necrosis, haemorrhage, oedema). All preoperative MR were evaluated by an experienced neuroradiologist and information on patients' demographics, clinical presentation, follow-up, and histopathology results study were collected. When available, preoperative CT examination was also evaluated. We found 14 e-GBs in 13 patients (9 males, 4 females, median age 63years) among 660 patients diagnosed with GB between 2010 and 2020. In 10 lesions, serial imaging revealed the transformation of e-GB in classic glioblastoma in a median time of 3months. Clinical presentation included stroke-like symptoms, vertigo, seizures and confusion. Preoperative plain CT was performed in 8/13 cases and in 7 e-GBs presented as a hyperdense lesion. Ten out of 14 lesions transformed in classic GB before surgical intervention or biopsy. All lesions revealed typical immunohistochemical pattern of primary glioblastoma. E-GB is a rare entity that can often lead to misdiagnosis. However, the radiologist should be aware of its imaging appearance to suggest the diagnosis and to request close imaging follow-up, hopefully improving the prognosis of this very aggressive disease.

Development of a Machine Learning Classifier Based on Radiomic Features Extracted From Post-Contrast 3D T1-Weighted MR Images to Distinguish Glioblastoma From Solitary Brain Metastasis.

ObjectivesTo differentiate Glioblastomas (GBM) and Brain Metastases (BM) using a radiomic features-based Machine Learning (ML) classifier trained from post-contrast three-dimensional T1-weighted (post-contrast 3DT1) MR imaging, and compare its performance in medical diagnosis versus human experts, on a testing cohort.MethodsWe enrolled 143 patients (71 GBM and 72 BM) in a retrospective bicentric study from January 2010 to May 2019 to train the classifier. Post-contrast 3DT1 MR images were performed on a 3-Tesla MR unit and 100 radiomic features were extracted. Selection and optimization of the Machine Learning (ML) classifier was performed using a nested cross-validation. Sensitivity, specificity, balanced accuracy, and area under the receiver operating characteristic curve (AUC) were calculated as performance metrics. The model final performance was cross-validated, then evaluated on a test set of 37 patients, and compared to human blind reading using a McNemar’s test.ResultsThe ML classifier had a mean [95% confidence interval] sensitivity of 85% [77; 94], a specificity of 87% [78; 97], a balanced accuracy of 86% [80; 92], and an AUC of 92% [87; 97] with cross-validation. Sensitivity, specificity, balanced accuracy and AUC were equal to 75, 86, 80 and 85% on the test set. Sphericity 3D radiomic index highlighted the highest coefficient in the logistic regression model. There were no statistical significant differences observed between the performance of the classifier and the experts’ blinded examination.ConclusionsThe proposed diagnostic support system based on radiomic features extracted from post-contrast 3DT1 MR images helps in differentiating solitary BM from GBM with high diagnosis performance and generalizability.

The success of machine learning algorithms developed with radiomic features obtained from preoperative contrast-enhanced MRI in the prediction of short-term survival in patients with glioblastoma

Purpose: This study aimed to evaluate the predictability of survival in patients with glioblastoma using a machine learning (ML) model developed with tissue analysis features obtained through preoperative post-contrast T1-weighted images(T1WI).&#x0D; Materials and Methods: The radiomic features of tumors were obtained from postcontrast T1WI of 60 glioblastoma patients. Radiomic properties, density, shape, and textural properties obtained from six matrices were included in the analysis. The patients' three- and six-month survival rates were recorded. Five different ML algorithms were applied to create predictive models [random forest, neural network, linear discriminant analysis(LDA), stochastic gradient descent (SGD), and support vector machine(SMV)]. &#x0D; Results: The mean survival time of the patients was 295.4 days, and the median value was 211.5 (17-1357) days. Among the models developed for three- and six-month survival prediction, the highest success was obtained from the LDA algorithm, in which the AUC values were calculated as 0.88 and 0.78, respectively.&#x0D; Conclusion: Using ML techniques, the success of predicting imaging-based patient survival was very high. With the development and widespread adoption of these techniques, ML models will be useful in deciding on treatment according to survival prediction in glioblastoma.

Clinical assessment of a biophysical model for distinguishing tumor progression from radiation necrosis

The efficacy of an imaging-driven mechanistic biophysical model of tumor growth for distinguishing radiation necrosis from tumor progression in patients with enhancing lesions following stereotactic radiosurgery (SRS) for brain metastasis is validated. We retrospectively assessed the model using 73 patients with 78 lesions and histologically confirmed radiation necrosis or tumor progression. Postcontrast T1-weighted MRI images were used to extract parameters for a mechanistic reaction-diffusion logistic growth model mechanically coupled to the surrounding tissue. The resulting model was then used to estimate mechanical stress fields, which were then compared with edema visualized on FLAIR imaging using DICE similarity coefficients. DICE, model, and standard radiographic morphometric analysis parameters were evaluated using a receiver operating characteristic (ROC) curve for prediction of radiation necrosis or tumor progression. Multivariate logistic regression models were then constructed using mechanistic model parameters or advanced radiomic features. An independent validation was performed to evaluate predictive performance. Tumor cell proliferation rate resulted in ROC AUC=0.86, 95% CI: 0.76-0.95, P<0.0001, 74% sensitivity and 95% specificity) and DICE similarity coefficient associated with high stresses demonstrated an ROC AUC=0.93, 95% CI: 0.86-0.99, P<0.0001, 81% sensitivity and 95% specificity. In a multivariate logistic regression model using an independent validation dataset, mechanistic modeling parameters had an ROC AUC of 0.95, with 94% sensitivity and 96% specificity. Imaging-driven biophysical modeling of tumor growth represents a novel method for accurately predicting clinically significant tumor behavior.

AUCseg: An Automatically Unsupervised Clustering Toolbox for 3D-Segmentation of High-Grade Gliomas in Multi-Parametric MR Images.

The segmentation of high-grade gliomas (HGG) using magnetic resonance imaging (MRI) data is clinically meaningful in neurosurgical practice, but a challenging task. Currently, most segmentation methods are supervised learning with labeled training sets. Although these methods work well in most cases, they typically require time-consuming manual labeling and pre-trained models. In this work, we propose an automatically unsupervised segmentation toolbox based on the clustering algorithm and morphological processing, named AUCseg. With our toolbox, the whole tumor was first extracted by clustering on T2-FLAIR images. Then, based on the mask acquired with whole tumor segmentation, the enhancing tumor was segmented on the post-contrast T1-weighted images (T1-CE) using clustering methods. Finally, the necrotic regions were segmented by morphological processing or clustering on T2-weighted images. Compared with K-means, Mini-batch K-means, and Fuzzy C Means (FCM), the Gaussian Mixture Model (GMM) clustering performs the best in our toolbox. We did a multi-sided evaluation of our toolbox in the BraTS2018 dataset and demonstrated that the whole tumor, tumor core, and enhancing tumor can be automatically segmented using default hyper-parameters with Dice score 0.8209, 0.7087, and 0.7254, respectively. The computing time of our toolbox for each case is around 22 seconds, which is at least 3 times faster than other state-of-the-art unsupervised methods. In addition, our toolbox has an option to perform semi-automatic segmentation via manually setup hyper-parameters, which could improve the segmentation performance. Our toolbox, AUCseg, is publicly available on Github. (https://github.com/Haifengtao/AUCseg).

MRI Highly Accelerated Wave-CAIPI T1-SPACE versus Standard T1-SPACE to detect brain gadolinium-enhancing lesions at 3T.

High-resolution three-dimensional (3D) post-contrast imaging of the brain is essential for comprehensive evaluation of inflammatory, neoplastic, and neurovascular diseases of the brain. 3D T1-weighted spin-echo-based sequences offer increased sensitivity for the detection of enhancing lesions but are relatively prolonged examinations. We evaluated whether a highly accelerated Wave-controlled aliasing in parallel imaging (Wave-CAIPI) post-contrast 3D T1-sampling perfection with application-optimized contrasts using different flip angle evolutions (T1-SPACE) sequence (Wave-T1-SPACE) was noninferior to the standard high-resolution 3D T1-SPACE sequence for visualizing enhancing lesions with comparable diagnostic quality. One hundred and three consecutive patients were prospectively evaluated with a standard post-contrast 3D T1-SPACE sequence (acquisition time [TA] = 4 min 19 s) and an optimized Wave-CAIPI 3D T1-SPACE sequence (TA = 1 min 40 s) that was nearly three times faster than the standard sequence. Two blinded neuroradiologists performed a head-to-head comparison to evaluate the visualization of enhancing pathology, perception of artifacts, and overall diagnostic quality. A 15% margin was used to test whether post-contrast Wave-T1-SPACE was noninferior to standard T1-SPACE. Wave-T1-SPACE was noninferior to standard T1-SPACE for delineating parenchymal and meningeal enhancing pathology (p<0.01). Wave-T1-SPACE showed marginally higher background noise compared to the standard sequence and was noninferior in the overall diagnostic quality (p = 0.03). Our findings show that Wave-T1-SPACE was noninferior to standard T1-SPACE for visualization of enhancing pathology and overall diagnostic quality with a three-fold reduction in acquisition time compared to the standard sequence. Wave-T1-SPACE may be used to accelerate 3D post-contrast T1-weighted spin-echo imaging without loss of clinically important information.

Development of a Fully Automated Glioma-Grading Pipeline Using Post-Contrast T1-Weighted Images Combined with Cloud-Based 3D Convolutional Neural Network

Glioma is the most common type of brain tumor, and its grade influences its treatment policy and prognosis. Therefore, artificial-intelligence-based tumor grading methods have been studied. However, in most studies, two-dimensional (2D) analysis and manual tumor-region extraction were performed. Additionally, deep learning research that uses medical images experiences difficulties in collecting image data and preparing hardware, thus hindering its widespread use. Therefore, we developed a 3D convolutional neural network (3D CNN) pipeline for realizing a fully automated glioma-grading system by using the pretrained Clara segmentation model provided by NVIDIA and our original classification model. In this method, the brain tumor region was extracted using the Clara segmentation model, and the volume of interest (VOI) created using this extracted region was assigned to a grading 3D CNN and classified as either grade II, III, or IV. Through evaluation using 46 regions, the grading accuracy of all tumors was 91.3%, which was comparable to that of the method using multi-sequence. The proposed pipeline scheme may enable the creation of a fully automated glioma-grading pipeline in a single sequence by combining the pretrained 3D CNN and our original 3D CNN.

Evaluating deep learning methods in detecting and segmenting different sizes of brain metastases on 3D post-contrast T1-weighted images.

Purpose: We investigate the impact of various deep-learning-based methods for detecting and segmenting metastases with different lesion volume sizes on 3D brain MR images. Approach: A 2.5D U-Net and a 3D U-Net were selected. We also evaluated weak learner fusion of the prediction features generated by the 2.5D and the 3D networks. A 3D fully convolutional one-stage (FCOS) detector was selected as a representative of bounding-box regression-based detection methods. A total of 422 3D post-contrast T1-weighted scans from patients with brain metastases were used. Performances were analyzed based on lesion volume, total metastatic volume per patient, and number of lesions per patient. Results: The performance of detection of the 2.5D and 3D U-Net methods had recall of and precision of for lesion volume but deteriorated as metastasis size decreased below to 0.58 to 0.74 in recall and 0.16 to 0.25 in precision. Compared the two U-Nets for detection capability, high precision was achieved by the 2.5D network, but high recall was achieved by the 3D network for all lesion sizes. The weak learner fusion achieved a balanced performance between the 2.5D and 3D U-Nets; particularly, it increased precision to 0.83 for lesion volumes of 0.1 to but decreased recall to 0.59. The 3D FCOS detector did not outperform the U-Net methods in detecting either the small or large metastases presumably because of the limited data size. Conclusions: Our study provides the performances of four deep learning methods in relationship to lesion size, total metastasis volume, and number of lesions per patient, providing insight into further development of the deep learning networks.

Feasibility of Simulated Postcontrast MRI of Glioblastomas and Lower-Grade Gliomas by Using Three-dimensional Fully Convolutional Neural Networks.

To evaluate the feasibility and accuracy of simulated postcontrast T1-weighted brain MR images generated by using precontrast MR images in patients with brain glioma. In this retrospective study, a three-dimensional deep convolutional neural network was developed to simulate T1-weighted postcontrast images from eight precontrast sequences in 400 patients (mean age, 57 years; 239 men; from 2015 to 2020), including 332 with glioblastoma and 68 with lower-grade gliomas. Performance was evaluated by using quantitative image similarity and error metrics and enhancing tumor overlap analysis. Performance was also assessed on a multicenter external dataset (n = 286 from the 2019 Multimodal Brain Tumor Segmentation Challenge; mean age, 60 years; ratio of men to women unknown) by using transfer learning. A subset of cases was reviewed by neuroradiologist readers to assess whether simulated images affected the ability to determine the tumor grade. Simulated whole-brain postcontrast images were both qualitatively and quantitatively similar to the real postcontrast images in terms of quantitative image similarity (structural similarity index of 0.84 ± 0.05), pixelwise error (symmetric mean absolute percent error of 3.65%), and enhancing tumor compartment overlap (Dice coefficient, 0.65 ± 0.25). Similar results were achieved with the external dataset (Dice coefficient, 0.62 ± 0.27). There was no difference in the ability of the neuroradiologist readers to determine the tumor grade in real versus simulated images (accuracy, 87.7% vs 90.6%; P = .87). The developed model was capable of producing simulated postcontrast T1-weighted MR images that were similar to real acquired images as determined by both quantitative analysis and radiologist assessment.Keywords: MR-Contrast Agent, MR-Imaging, CNS, Brain/Brain Stem, Contrast Agents-Intravenous, Neoplasms-Primary, Experimental Investigations, Technology Assessment, Supervised Learning, Transfer Learning, Convolutional Neural Network, Deep Learning Algorithms, Machine Learning Algorithms Supplemental material is available for this article. © RSNA, 2021.

Leptomeningeal enhancement on post-contrast FLAIR images for early diagnosis of Susac syndrome

Background: Leptomeningeal enhancement (LME) is a key feature of Susac syndrome (SuS) but is only occasionally depicted on post-contrast T1-weighted images (T1-WI). Objective: As post-contrast fluid-attenuated inversion recovery (FLAIR) may be more sensitive, our aim was to assess LME in SuS on this sequence. Methods: From 2010 to 2020, 20 patients with definite SuS diagnosis were retrospectively enrolled in this multicentre study. Two radiologists independently assessed the number of LME on post-contrast FLAIR and T1-WI acquisitions performed before any treatment. A chi-square test was used to compare both sequences and the interrater agreement was calculated. Results: Thirty-five magnetic resonance imagings (MRIs) were performed before treatment, including 19 post-contrast FLAIR images in 17 patients and 25 post-contrast T1-WI in 19 patients. In terms of patients, LME was observed on all post-contrast FLAIR, contrary to post-contrast T1-WI (17/17 (100%) vs. 15/19 (79%), p < 0.05). In terms of sequences, LME was observed on all post-contrast FLAIR, contrary to post-contrast T1-WI (19/19 (100%) vs. 16/25 (64%), p < 0.005). LME was disseminated at both supratentorial (19/19) and infratentorial (18/19) levels on post-contrast FLAIR, contrary to post-contrast T1-WI (3/25 and 9/25, respectively). Interrater agreement was excellent for post-contrast FLAIR (κ = 0.95) but only moderate for post-contrast T1-WI (κ = 0.61). Conclusion: LME was always observed and easily visible on post-contrast FLAIR images prior to SuS treatment. In association with other MRI features, it is highly indicative of SuS.

Generative Adversarial Networks to Synthesize Missing T1 and FLAIR MRI Sequences for Use in a Multisequence Brain Tumor Segmentation Model.

Background Missing MRI sequences represent an obstacle in the development and use of deep learning (DL) models that require multiple inputs. Purpose To determine if synthesizing brain MRI scans using generative adversarial networks (GANs) allows for the use of a DL model for brain lesion segmentation that requires T1-weighted images, postcontrast T1-weighted images, fluid-attenuated inversion recovery (FLAIR) images, and T2-weighted images. Materials and Methods In this retrospective study, brain MRI scans obtained between 2011 and 2019 were collected, and scenarios were simulated in which the T1-weighted images and FLAIR images were missing. Two GANs were trained, validated, and tested using 210 glioblastomas (GBMs) (Multimodal Brain Tumor Image Segmentation Benchmark [BRATS] 2017) to generate T1-weighted images from postcontrast T1-weighted images and FLAIR images from T2-weighted images. The quality of the generated images was evaluated with mean squared error (MSE) and the structural similarity index (SSI). The segmentations obtained with the generated scans were compared with those obtained with the original MRI scans using the dice similarity coefficient (DSC). The GANs were validated on sets of GBMs and central nervous system lymphomas from the authors' institution to assess their generalizability. Statistical analysis was performed using the Mann-Whitney, Friedman, and Dunn tests. Results Two hundred ten GBMs from the BRATS data set and 46 GBMs (mean patient age, 58 years ± 11 [standard deviation]; 27 men [59%] and 19 women [41%]) and 21 central nervous system lymphomas (mean patient age, 67 years ± 13; 12 men [57%] and nine women [43%]) from the authors' institution were evaluated. The median MSE for the generated T1-weighted images ranged from 0.005 to 0.013, and the median MSE for the generated FLAIR images ranged from 0.004 to 0.103. The median SSI ranged from 0.82 to 0.92 for the generated T1-weighted images and from 0.76 to 0.92 for the generated FLAIR images. The median DSCs for the segmentation of the whole lesion, the FLAIR hyperintensities, and the contrast-enhanced areas using the generated scans were 0.82, 0.71, and 0.92, respectively, when replacing both T1-weighted and FLAIR images; 0.84, 0.74, and 0.97 when replacing only the FLAIR images; and 0.97, 0.95, and 0.92 when replacing only the T1-weighted images. Conclusion Brain MRI scans generated using generative adversarial networks can be used as deep learning model inputs in case MRI sequences are missing. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Zhong in this issue. An earlier incorrect version of this article appeared online. This article was corrected on April 12, 2021.

Quantitative analysis of gadoxetic acid-enhanced MRI for the differential diagnosis of focal liver lesions: Comparison between estimated intralesional gadoxetic acid retention by T1 mapping and conventional processing methods

PurposeTo compare the estimated quantity of intratumor gadoxetic acid retention using T1 mapping of gadoxetic acid-enhanced magnetic resonance imaging (MRI) versus conventional processing methods for the differential diagnosis of focal liver lesions. MethodsSeventy patients with hepatic lesions (colorectal metastasis (CRM) [n = 28], hepatocellular carcinoma (HCC) [n = 20], hemangioma [n = 12], and intrahepatic cholangiocarcinoma (ICC) [n = 10]) underwent gadoxetic acid-enhanced MRI, including pre- and post-contrast T1-weighted imaging and T1 mapping. Quantitative analyses included the lesion-to-liver signal intensity ratio (SIR) on hepatobiliary phase images, the pre- and post-contrast lesion T1 value difference (ΔT1 [ms]), and the lesion retention index (LRI [%]), which was the estimated intralesional gadoxetic acid retention calculated on pre- and post-contrast T1 maps using a two-compartment pharmacokinetic model. Results were compared between the four subcategories of focal liver lesions using the Kruskal-Wallis test, followed by the post-hoc Dunn’s test and receiver operating characteristic (ROC) analysis to distinguish between pairs of the four lesion subcategories. ResultsThis study identified significant differences in the LRI of the four lesion subcategories (p < 0.01), without significant differences in ΔT1 or SIR. Post-hoc analysis demonstrated significant differences in CRM vs. hemangioma (p < 0.01), hemangioma vs. ICC (p < 0.01), and HCC vs. ICC (p = 0.047) for the LRI. ConclusionsThe quantity of intratumor gadoxetic acid retention estimated using pre- and post- contrast T1 mapping could distinguish focal liver lesions, unlike conventional processing methods, and captured unique lesion characteristics.

Abstract PO-023: Differentiation of tumefactive multiple sclerosis and glioblastoma using radiomics features extracted from magnetic resonance imaging and machine learning

Abstract Introduction Tumefactive multiple sclerosis (MS) and glioblastoma (GBM) can be difficult to differentiate. Clinical presentations are nonspecific and may occur in patients with both diseases. Magnetic resonance imaging (MRI) scans are routinely used to generate an initial differential diagnosis. Although imaging features can help distinguish these entities, there are no pathognomonic signs, neither on traditional nor advanced MRI. Moreover, the ability to differentiate these entities depends on the radiologists’ experience. Ensuring the correct diagnosis has important therapeutic and prognostic implications, as the management of these diseases is very different. To date, a brain biopsy often represents the only option to obtain a diagnosis, but it may be inconclusive and could add unnecessary morbidity. We hypothesize that a machine-learning tool based on routinely performed MRI scans could guarantee a more reliable, fast, and reproducible diagnosis. Methods Only patients with a histological diagnosis were included in the study. Patients that underwent any form of radiation therapy before the surgery were excluded to avoid confounding factors due to treatment-related effects. A total of 99 MS and 257 GBMs were included, and preoperative T2-weighted images were collected. For each patient, the lesions' segmentation was performed, including both the contrast-enhancing and non-enhancing areas of the lesions. The segmentation masks were used to extract radiomic features from the T2-weighted sequences. All MRI images were standardized to the same mean and standard deviation before feature extraction and resampled to a voxel size of 1.0 × 1.0 × 1.0 mm. Features extraction was performed using Pyradiomics (v. 3.0.1) and included first order, shape, and texture features, totaling 100 features. The classifier consisted of a support vector machine (SVM) model. Patients were randomly assigned to the training and test sets with an 80-20% proportion for each class. To select the best hyperparameters for the SVM classifier, a grid-search analysis was performed on the training set using a nested cross-validation scheme (k=5). All experiments were performed using Python (v. 3.5) and the Scikit-learn library (v. 0.23.1). Results The mean accuracy and area under the ROC curve for the best set of hyperparameters on the validation set were 0.83 and 0.86, respectively. The weighted-average precision, recall, and F1 scores on the test set were 0.78, 0.76, and 0.77, respectively. Conclusions Radiomic features extracted from a routinely acquired MRI sequence showed promising results in differentiating tumefactive multiple sclerosis from glioblastoma, opening the possibility to obtain a reliable diagnosis without the need for a brain biopsy. Future experiments Future experiments will focus on using post-contrast T1-weighted images for the radiomic features extractions, testing multiple machine learning models and features selection algorithms, and including more diagnostic classes (i.e., metastasis, lymphomas). Citation Format: Gian Marco Conte, Jeanette E. Eckel-Passow, W. Oliver Tobin, Paul Decker, Daniel H. Lachance, Robert B. Jenkins, Bradley J. Erickson. Differentiation of tumefactive multiple sclerosis and glioblastoma using radiomics features extracted from magnetic resonance imaging and machine learning [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-023.

Diagnostic accuracy of MRI textural analysis in the classification of breast tumors

ObjectiveTo investigate whether textural analysis (TA) of MRI heterogeneity may play a role in the clinical assessment and classification of breast tumors. Materials and methodsFor this retrospective study, patients with breast masses ≥1 cm on contrast-enhanced MRI were obtained in 69 women (mean age: 51 years; range 21–78 years) with 77 masses (38 benign, 39 malignant) from 2006 to 2018. The selected single slice sagittal peak post-contrast T1-weighted image was analyzed with commercially available TA software [TexRAD Ltd., UK]. Eight histogram TA parameters were evaluated at various spatial scaling factors (SSF) including mean pixel intensity, standard deviation of the pixel histogram (SD), entropy, mean of the positive pixels (MPP), skewness, kurtosis, sigma, and Tx_sigma. Additional statistical tests were used to determine their predictiveness. ResultsEntropy showed a significant difference between benign and malignant tumors at all textural scales (p < 0.0001) and kurtosis was significant at SSF = 0–5 (p = 0.0026–0.0241). The single best predictor was entropy at SSF = 4 with AUC = 0.80, giving a sensitivity of 95% and specificity of 53%. An AUC of 0.91 was found using a model combining entropy with sigma, which yielded better performance with a sensitivity of 92% and specificity of 79%. ConclusionTA of breast masses has the potential to assist radiologists in categorizing tumors as benign or malignant on MRI. Measurements of entropy, kurtosis, and entropy combined with sigma may provide the best predictability.