Dataset Information

Prediction of IDH genotype in gliomas with dynamic susceptibility contrast perfusion MR imaging using an explainable recurrent neural network.

ABSTRACT: BACKGROUND:The aim of this study was to predict isocitrate dehydrogenase (IDH) genotypes of gliomas using an interpretable deep learning application for dynamic susceptibility contrast (DSC) perfusion MRI. METHODS:Four hundred sixty-three patients with gliomas who underwent preoperative MRI were enrolled in the study. All the patients had immunohistopathologic diagnoses of either IDH-wildtype or IDH-mutant gliomas. Tumor subregions were segmented using a convolutional neural network followed by manual correction. DSC perfusion MRI was performed to obtain T2* susceptibility signal intensity-time curves from each subregion of the tumors: enhancing tumor, non-enhancing tumor, peritumoral edema, and whole tumor. These, with arterial input functions, were fed into a neural network as multidimensional inputs. A convolutional long short-term memory model with an attention mechanism was developed to predict IDH genotypes. Receiver operating characteristics analysis was performed to evaluate the model. RESULTS:The IDH genotype predictions had an accuracy, sensitivity, and specificity of 92.8%, 92.6%, and 93.1%, respectively, in the validation set (area under the curve [AUC], 0.98; 95% confidence interval [CI], 0.969-0.991) and 91.7%, 92.1%, and 91.5%, respectively, in the test set (AUC, 0.95; 95% CI, 0.898-0.982). In temporal feature analysis, T2* susceptibility signal intensity-time curves obtained from DSC perfusion MRI with attention weights demonstrated high attention on the combination of the end of the pre-contrast baseline, up/downslopes of signal drops, and/or post-bolus plateaus for the curves used to predict IDH genotype. CONCLUSIONS:We developed an explainable recurrent neural network model based on DSC perfusion MRI to predict IDH genotypes in gliomas.

SUBMITTER: Choi KS

PROVIDER: S-EPMC7594560 | biostudies-literature | 2019 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Prediction of IDH genotype in gliomas with dynamic susceptibility contrast perfusion MR imaging using an explainable recurrent neural network.

Choi Kyu Sung KS Choi Seung Hong SH Jeong Bumseok B

Neuro-oncology 20190901 9

<h4>Background</h4>The aim of this study was to predict isocitrate dehydrogenase (IDH) genotypes of gliomas using an interpretable deep learning application for dynamic susceptibility contrast (DSC) perfusion MRI.<h4>Methods</h4>Four hundred sixty-three patients with gliomas who underwent preoperative MRI were enrolled in the study. All the patients had immunohistopathologic diagnoses of either IDH-wildtype or IDH-mutant gliomas. Tumor subregions were segmented using a convolutional neural netwo ...[more]

PMID: 31127834

Similar Datasets

Project description:PurposeThe purpose of this study was to evaluate the accuracy of diffusion-weighted imaging (DWI) and dynamic contrast-enhanced (DCE) perfusion MR imaging for distinguishing tumor recurrence from post-treatment effect as alternatives to dynamic-susceptibility contrast-enhanced (DSC) perfusion MR imaging when the DSC image is uninterpretable.Materials and methodsThis retrospective study was approved by our institutional review board. Seventy one post-treatment glioblastoma patients who showed enlarged contrast-enhancing lesions on follow-up MR images after concurrent chemoradiotherapy and uninterpretable DSC images for corresponding enhancing lesions, underwent additional DWI and DCE MR imaging. The primary outcome was the frequency of interpretable DWI and DCE MR cases in these 71 patients. The secondary outcome was the area under the receiver operating characteristic curve (AUC) of DWI and DCE imaging parameters for distinguishing tumor recurrence from post-treatment effect in selected patients with interpretable DWI and DCE images. The imaging parameters were quantified as 10% cumulative histogram cutoff of apparent diffusion coefficient (ADC10) and 90% cumulative histogram cutoff of initial area under the time signal intensity curve (IAUC90). The AUCs were cross-validated by using leave-one-out method.ResultsOf the 71 patients, the uninterpretable DSC images were associated with treatment-related hemorrhage within the corresponding enhancing lesions (n = 54, 76.1%) and a near skull base location (n = 17, 23.9%). The frequencies of interpretable DWI and DCE image were 51 (71.8%) and 59 (83.1%) of the 71 cases with uninterpretable DSC images, respectively. Of the 45 selected patients with interpretable DWI and DCE images, the combination of DWI with DCE imaging showed a superior diagnostic performance than DWI or DCE imaging alone for differentiating tumor recurrence from post-treatment effect (cross-validated AUC: 0.78 versus 0.55 and 0.73 for reader 1; cross-validated AUC: 0.78 versus 0.53 and 0.75 for reader 2, respectively). Cross-validated accuracy of the single and combined imaging parameters also showed the highest for the combination of DWI with DCE MR imaging (72.9% for reader 1; 72.5% for reader 2) and the lowest for DWI alone (54.0% for reader 1; 56.4% for reader 2). Inter-reader agreement for DCE imaging was higher than that for DWI (intraclass correlation coefficient: 0.95 versus 0.87).ConclusionDCE MR imaging could be a superior and more reproducible imaging biomarker than DWI for differentiating tumor recurrence from post-treatment effect in patients with post-treatment glioblastoma when DSC MR images are not interpretable.

Project description:ObjectivesCerebral blood flow (CBF) measurements after endovascular therapy (EVT) for acute ischemic stroke are important to distinguish early secondary injury related to persisting ischemia from that related to reperfusion when considering clinical response and infarct growth.MethodsWe compare reperfusion quantified by the modified Thrombolysis in Cerebral Infarction Score (mTICI) with perfusion measured by MRI dynamic contrast-enhanced perfusion within 5 h of EVT anterior circulation stroke. MR perfusion (rCBF, rCBV, rTmax, rT0) and mTICI scores were included in a predictive model for change in NIHSS at 24 h and diffusion-weighted imaging (DWI) lesion growth (acute to 24 h MRI) using a machine learning RRELIEFF feature selection coupled with a support vector regression.ResultsFor all perfusion parameters, mean values within the acute infarct for the TICI-2b group (considered clinically good reperfusion) were not significantly different from those in the mTICI <2b (clinically poor reperfusion). However, there was a statistically significant difference in perfusion values within the acute infarct region of interest between the mTICI-3 group versus both mTICI-2b and <2b (p = 0.02). The features that made up the best predictive model for change in NIHSS and absolute DWI lesion volume change was rT0 within acute infarct ROI and admission CTA collaterals respectively. No other variables, including mTICI scores, were selected for these best models. The correlation coefficients (Root mean squared error) for the cross-validation were 0.47 (13.7) and 0.51 (5.7) for change in NIHSS and absolute DWI lesion volume change.ConclusionMR perfusion following EVT provides accurate physiological approach to understanding the relationship of CBF, clinical outcome, and DWI growth.Advances in knowledgeMR perfusion CBF acquired is a robust, objective reperfusion measurement providing following recanalization of the target occlusion which is critical to distinguish potential therapeutic harm from the failed technical success of EVT as well as improve the responsiveness of clinical trial outcomes to disease modification.

Project description:MR Fingerprinting (MRF)-based Arterial-Spin-Labeling (ASL) has the potential to measure multiple parameters such as cerebral blood flow (CBF), bolus arrival time (BAT), and tissue T1 in a single scan. However, the previous reports have only demonstrated a proof-of-principle of the technique but have not examined the performance of the sequence in the context of key imaging parameters. Furthermore, there has not been a study to directly compare the technique to clinically used perfusion method of dynamic-susceptibility-contrast (DSC) MRI. The present report consists of two studies. In the first study (N = 8), we examined the dependence of MRF-ASL sequence on TR time pattern. Ten different TR patterns with a range of temporal characteristics were examined by both simulations and experiments. The results revealed that there was a significance dependence of the sequence performance on TR pattern (p < 0.001), although there was not a single pattern that provided dramatically improvements. Among the TR patterns tested, a sinusoidal pattern with a period of 125 TRs provided an overall best estimation in terms of spatial consistency. These experimental observations were consistent with those of numerical simulations. In the second study (N = 8), we compared MRF-ASL results with those of DSC MRI. It was found that MRF-ASL and DSC MRI provided highly comparable maps of cerebral blood flow (CBF) and bolus-arrival-time (BAT), with spatial correlation coefficients of 0.79 and 0.91, respectively. However, in terms of quantitative values, BAT obtained with MRF-ASL was considerably lower than that from DSC (p < 0.001), presumably because of the differences in tracer characteristics in terms of diffusible versus intravascular tracers. Test-retest assessment of MRF-ASL MRI revealed that the spatial correlations of parametric maps were 0.997, 0.962, 0.746 and 0.863 for B1 + , T1 , CBF, and BAT, respectively. MRF-ASL is a promising technique for assessing multiple perfusion parameters simultaneously without contrast agent.

Dataset Information

Prediction of IDH genotype in gliomas with dynamic susceptibility contrast perfusion MR imaging using an explainable recurrent neural network.

Publications

Prediction of IDH genotype in gliomas with dynamic susceptibility contrast perfusion MR imaging using an explainable recurrent neural network.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure