A Two-Stage Training Framework Using Multicontrast MRI Radiomics for IDH Mutation Status Prediction in Glioma.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To develop a radiomics framework for preoperative MRI-based prediction of IDH mutation status, a crucial glioma prognostic indicator. Materials and Methods Radiomics features (shape, first-order statistics, and texture) were extracted from the whole tumor or the combination of nonenhancing, necrosis, and edema regions. Segmentation masks were obtained via the federated tumor segmentation tool or the original data source. Boruta, a wrapper-based feature selection algorithm, identified relevant features. Addressing the imbalance between mutated and wild-type cases, multiple prediction models were trained on balanced data subsets using Random Forest or XGBoost and assembled to build the final classifier. The framework was evaluated using retrospective MRI scans from three public datasets (The Cancer Imaging Archive (TCIA, 227 patients), the University of California San Francisco Preoperative Diffuse Glioma MRI dataset (UCSF, 495 patients), and the Erasmus Glioma Database (EGD, 456 patients)) and internal datasets collected from UT Southwestern Medical Center (UTSW, 356 patients), New York University (NYU, 136 patients), and University of Wisconsin-Madison (UWM, 174 patients). TCIA and UTSW served as separate training sets, while the remaining data constituted the test set (1617 or 1488 testing cases, respectively). Results The best-performing models trained on the TCIA dataset achieved area under the receiver operating characteristic curve (AUC) values of 0.89 for UTSW, 0.86 for NYU, 0.93 for UWM, 0.94 for UCSF, and 0.88 for EGD test sets. The best-performing models trained on the UTSW dataset achieved slightly higher AUCs: 0.92 for TCIA, 0.88 for NYU, 0.96 for UWM, 0.93 for UCSF, and 0.90 for EGD. Conclusion This MRI radiomics-based framework shows promise for accurate preoperative prediction of IDH mutation status in patients with glioma. Published under a CC BY 4.0 license.

Synthesizing 3D Multi-Contrast Brain Tumor MRIs Using Tumor Mask Conditioning.

Data scarcity and data imbalance are two major challenges in training deep learning models on medical images, such as brain tumor MRI data. The recent advancements in generative artificial intelligence have opened new possibilities for synthetically generating MRI data, including brain tumor MRI scans. This approach can be a potential solution to mitigate the data scarcity problem and enhance training data availability. This work focused on adapting the 2D latent diffusion models to generate 3D multi-contrast brain tumor MRI data with a tumor mask as the condition. The framework comprises two components: a 3D autoencoder model for perceptual compression and a conditional 3D Diffusion Probabilistic Model (DPM) for generating high-quality and diverse multi-contrast brain tumor MRI samples, guided by a conditional tumor mask. Unlike existing works that focused on generating either 2D multi-contrast or 3D single-contrast MRI samples, our models generate multi-contrast 3D MRI samples. We also integrated a conditional module within the UNet backbone of the DPM to capture the semantic class-dependent data distribution driven by the provided tumor mask to generate MRI brain tumor samples based on a specific brain tumor mask. We trained our models using two brain tumor datasets: The Cancer Genome Atlas (TCGA) public dataset and an internal dataset from the University of Texas Southwestern Medical Center (UTSW). The models were able to generate high-quality 3D multi-contrast brain tumor MRI samples with the tumor location aligned by the input condition mask. The quality of the generated images was evaluated using the Fréchet Inception Distance (FID) score. This work has the potential to mitigate the scarcity of brain tumor data and improve the performance of deep learning models involving brain tumor MRI data.

Single-Fraction Versus Fractionated Preoperative Radiosurgery for Resected Brain Metastases: A PROPS-BM International Multicenter Cohort Study.

PURPOSE: Preoperative stereotactic radiosurgery (SRS) is a feasible alternative to postoperative SRS for resected brain metastases (BM). Most reported studies of preoperative SRS used single-fraction SRS (SF-SRS). The goal of this study was to compare outcomes and toxicity of preoperative SF-SRS with multifraction (3-5 fractions) SRS (MF-SRS) in a large international multicenter cohort (Preoperative Radiosurgery for Brain Metastases-PROPS-BM). METHODS AND MATERIALS: Patients with BM from solid cancers, of which at least 1 lesion was treated with preoperative SRS followed by planned resection, were included from 8 institutions. SRS to synchronous intact BM was allowed. Exclusion criteria included prior or planned whole brain radiation therapy. Intracranial outcomes were estimated using cumulative incidence with competing risk of death. Propensity score matched (PSM) analyses were performed. RESULTS: The study cohort included 404 patients with 416 resected index lesions, of which SF-SRS and MF-SRS were used for 317 (78.5%) and 87 patients (21.5%), respectively. Median dose was 15 Gy in 1 fraction for SF-SRS and 24 Gy in 3 fractions for MF-SRS. Univariable analysis demonstrated that SF-SRS was associated with higher cavity local recurrence (LR) compared with MF-SRS (2-year: 16.3% vs 2.9%; P = .004), which was also demonstrated in multivariable analysis. PSM yielded 81 matched pairs (n = 162). PSM analysis also demonstrated significantly higher rate of cavity LR with SF-SRS (2-year: 19.8% vs 3.3%; P = .003). There was no difference in adverse radiation effect, meningeal disease, or overall survival between cohorts in either analysis. CONCLUSIONS: Preoperative MF-SRS was associated with significantly reduced risk of cavity LR in both the unmatched and PSM analyses. There was no difference in adverse radiation effect, meningeal disease, or overall survival based on fractionation. MF-SRS may be a preferred option for neoadjuvant radiation therapy of resected BMs. Additional confirmatory studies are needed. A phase 3 randomized trial of single-fraction preoperative versus postoperative SRS (NRG-BN012) is ongoing (NCT05438212).

Rationale and design of the CONFIRM2 (Quantitative COroNary CT Angiography Evaluation For Evaluation of Clinical Outcomes: An InteRnational, Multicenter Registry) study.

BACKGROUND: In the last 15 years, large registries and several randomized clinical trials have demonstrated the diagnostic and prognostic value of coronary computed tomography angiography (CCTA). Advances in CT scanner technology and developments of analytic tools now enable accurate quantification of coronary artery disease (CAD), including total coronary plaque volume and low attenuation plaque volume. The primary aim of CONFIRM2, (Quantitative COroNary CT Angiography Evaluation For Evaluation of Clinical Outcomes: An InteRnational, Multicenter Registry) is to perform comprehensive quantification of CCTA findings, including coronary, non-coronary cardiac, non-cardiac vascular, non-cardiac findings, and relate them to clinical variables and cardiovascular clinical outcomes. DESIGN: CONFIRM2 is a multicenter, international observational cohort study designed to evaluate multidimensional associations between quantitative phenotype of cardiovascular disease and future adverse clinical outcomes in subjects undergoing clinically indicated CCTA. The targeted population is heterogenous and includes patients undergoing CCTA for atherosclerotic evaluation, valvular heart disease, congenital heart disease or pre-procedural evaluation. Automated software will be utilized for quantification of coronary plaque, stenosis, vascular morphology and cardiac structures for rapid and reproducible tissue characterization. Up to 30,000 patients will be included from up to 50 international multi-continental clinical CCTA sites and followed for 3-4 years. SUMMARY: CONFIRM2 is one of the largest CCTA studies to establish the clinical value of a multiparametric approach to quantify the phenotype of cardiovascular disease by CCTA using automated imaging solutions.

MRI-Based Deep Learning Method for Classification of IDH Mutation Status.

Isocitrate dehydrogenase (IDH) mutation status has emerged as an important prognostic marker in gliomas. This study sought to develop deep learning networks for non-invasive IDH classification using T2w MR images while comparing their performance to a multi-contrast network. Methods: Multi-contrast brain tumor MRI and genomic data were obtained from The Cancer Imaging Archive (TCIA) and The Erasmus Glioma Database (EGD). Two separate 2D networks were developed using nnU-Net, a T2w-image-only network (T2-net) and a multi-contrast network (MC-net). Each network was separately trained using TCIA (227 subjects) or TCIA + EGD data (683 subjects combined). The networks were trained to classify IDH mutation status and implement single-label tumor segmentation simultaneously. The trained networks were tested on over 1100 held-out datasets including 360 cases from UT Southwestern Medical Center, 136 cases from New York University, 175 cases from the University of Wisconsin-Madison, 456 cases from EGD (for the TCIA-trained network), and 495 cases from the University of California, San Francisco public database. A receiver operating characteristic curve (ROC) was drawn to calculate the AUC value to determine classifier performance. Results: T2-net trained on TCIA and TCIA + EGD datasets achieved an overall accuracy of 85.4% and 87.6% with AUCs of 0.86 and 0.89, respectively. MC-net trained on TCIA and TCIA + EGD datasets achieved an overall accuracy of 91.0% and 92.8% with AUCs of 0.94 and 0.96, respectively. We developed reliable, high-performing deep learning algorithms for IDH classification using both a T2-image-only and a multi-contrast approach. The networks were tested on more than 1100 subjects from diverse databases, making this the largest study on image-based IDH classification to date.

Modeling gamma knife radiosurgical toxicity for multiple brain metastases.

BACKGROUND AND PURPOSE: Radiation oncology protocols for single fraction radiosurgery recommend setting dosing criteria based on assumed risk of radionecrosis, which can be predicted by the 12 Gy normal brain volume (V12). In this study, we show that tumor surface area (SA) and a simple power-law model using only preplan variables can estimate and minimize radiosurgical toxicity. MATERIALS AND METHODS: A 245-patient cohort with 1217 brain metastases treated with single or distributed Gamma Knife sessions was reviewed retrospectively. Univariate and multivariable linear regression models and power-law models determined which modeling parameters best predicted V12. The V12 power-law model, represented by a product of normalized Rx dose Rxn, and tumor longest axial dimension LAD (V12 âˆ¼ Rxn1.5*LAD2), was independently validated using a secondary 63-patient cohort with 302 brain metastases. RESULTS: Surface area was the best univariate linear predictor of V12 (adjR2 = 0.770), followed by longest axial dimension (adjR2 = 0.755) and volume (adjR2 = 0.745). The power-law model accounted for 90% variance in V12 for 1217 metastatic lesions (adjR2 = 0.906) and 245 patients (adjR2 = 0.896). The average difference ΔV12 between predicted and measured V12s was (0.28 ± 0.55) cm3 per lesion and (1.0 ± 1.2) cm3 per patient. The power-law predictive capability was validated using a secondary 63-patient dataset (adjR2 = 0.867) with 302 brain metastases (adjR2 = 0.825). CONCLUSION: Surface area was the most accurate univariate predictor of V12 for metastatic lesions. We developed a preplan model for brain metastases that can help better estimate radionecrosis risk, determine prescription doses given a target V12, and provide safe dose escalation strategies without the use of any planning software.

Risk Factors for Progression and Toxic Effects After Preoperative Stereotactic Radiosurgery for Patients With Resected Brain Metastases.

Importance: Preoperative stereotactic radiosurgery (SRS) has been demonstrated as a feasible alternative to postoperative SRS for resectable brain metastases (BMs) with potential benefits in adverse radiation effects (AREs) and meningeal disease (MD). However, mature large-cohort multicenter data are lacking. Objective: To evaluate preoperative SRS outcomes and prognostic factors from a large international multicenter cohort (Preoperative Radiosurgery for Brain Metastases-PROPS-BM). Design, Setting, and Participants: This multicenter cohort study included patients with BMs from solid cancers, of which at least 1 lesion received preoperative SRS and a planned resection, from 8 institutions. Radiosurgery to synchronous intact BMs was allowed. Exclusion criteria included prior or planned whole-brain radiotherapy and no cranial imaging follow-up. Patients were treated between 2005 and 2021, with most treated between 2017 and 2021. Exposures: Preoperative SRS to a median dose to 15 Gy in 1 fraction or 24 Gy in 3 fractions delivered at a median (IQR) of 2 (1-4) days before resection. Main Outcomes and Measures: The primary end points were cavity local recurrence (LR), MD, ARE, overall survival (OS), and multivariable analysis of prognostic factors associated with these outcomes. Results: The study cohort included 404 patients (214 women [53%]; median [IQR] age, 60.6 [54.0-69.6] years) with 416 resected index lesions. The 2-year cavity LR rate was 13.7%. Systemic disease status, extent of resection, SRS fractionation, type of surgery (piecemeal vs en bloc), and primary tumor type were associated with cavity LR risk. The 2-year MD rate was 5.8%, with extent of resection, primary tumor type, and posterior fossa location being associated with MD risk. The 2-year any-grade ARE rate was 7.4%, with target margin expansion greater than 1 mm and melanoma primary being associated with ARE risk. Median OS was 17.2 months (95% CI, 14.1-21.3 months), with systemic disease status, extent of resection, and primary tumor type being the strongest prognostic factors associated with OS. Conclusions and Relevance: In this cohort study, the rates of cavity LR, ARE, and MD after preoperative SRS were found to be notably low. Several tumor and treatment factors were identified that are associated with risk of cavity LR, ARE, MD, and OS after treatment with preoperative SRS. A phase 3 randomized clinical trial of preoperative vs postoperative SRS (NRG BN012) has began enrolling (NCT05438212).

Errata: Brain tumor IDH, 1p/19q, and MGMT molecular classification using MRI-based deep learning: an initial study on the effect of motion and motion correction.

[This corrects the article DOI: 10.1117/1.JMI.9.1.016001.].

Corrigendum to: A novel fully automated MRI-based deep-learning method for classification of 1p/19q co-deletion status in brain gliomas.

[This corrects the article DOI: 10.1093/noajnl/vdaa066.].

NCCN Guidelines® Insights: Central Nervous System Cancers, Version 2.2022.

The NCCN Guidelines for Central Nervous System (CNS) Cancers focus on management of the following adult CNS cancers: glioma (WHO grade 1, WHO grade 2-3 oligodendroglioma [1p19q codeleted, IDH-mutant], WHO grade 2-4 IDH-mutant astrocytoma, WHO grade 4 glioblastoma), intracranial and spinal ependymomas, medulloblastoma, limited and extensive brain metastases, leptomeningeal metastases, non-AIDS-related primary CNS lymphomas, metastatic spine tumors, meningiomas, and primary spinal cord tumors. The information contained in the algorithms and principles of management sections in the NCCN Guidelines for CNS Cancers are designed to help clinicians navigate through the complex management of patients with CNS tumors. Several important principles guide surgical management and treatment with radiotherapy and systemic therapy for adults with brain tumors. The NCCN CNS Cancers Panel meets at least annually to review comments from reviewers within their institutions, examine relevant new data from publications and abstracts, and reevaluate and update their recommendations. These NCCN Guidelines Insights summarize the panel's most recent recommendations regarding molecular profiling of gliomas.

Impact of CDKN2A/B, MTAP, and TERT Genetic Alterations on Survival in IDH Wild Type Glioblastomas.

PURPOSE: Poor outcomes in IDH wild-type (IDHwt) glioblastomas indicate the need to determine which genetic alterations can indicate poor survival and guidance of patient specific treatment options. We sought to identify the genetic alterations in these patients that predict for survival when adjusting particularly for treatments and other genetic alterations. METHODS: A cohort of 167 patients with pathologically confirmed IDHwt glioblastomas treated at our institution was retrospectively reviewed. Next generation sequencing was performed for each patient to determine tumor genetic alterations. Multivariable cox proportional hazards analysis for overall survival (OS) was performed to control for patient variables. RESULTS: CDKN2A, CDKN2B, and MTAP deletion predict for worse OS independently of other genetic alterations and patient characteristics (hazard ratio [HR] 2.192, p = 0.0017). Patients with CDKN2A copy loss (HR 2.963, p = 0.0037) or TERT mutated (HR 2.815, p = 0.0008) glioblastomas exhibited significant associations between radiation dose and OS, while CDKN2A and TERT wild type patients did not. CDKN2A deleted patients with NF1 mutations had worse OS (HR 1.990, p = 0.0540), while CDKN2A wild type patients had improved OS (HR 0.229, p = 0.0723). Patients with TERT mutated glioblastomas who were treated with radiation doses < 45 Gy (HR 3.019, p = 0.0010) but not those treated with ≥ 45 Gy exhibited worse OS compared to those without TERT mutations. CONCLUSION: In IDHwt glioblastomas, CDKN2A, CDKN2B, and MTAP predict for poor prognosis. TERT and CDKN2A mutations are associated with worse survival only when treated with lower radiation doses, thus potentially providing a genetic marker that can inform clinicians on proper dose-fractionation schemes.

Socioeconomic and demographic determinants of radiation treatment and outcomes in glioblastoma patients.

Introduction: Poor outcomes in glioblastoma patients, despite advancing treatment paradigms, indicate a need to determine non-physiologic prognostic indicators of patient outcome. The impact of specific socioeconomic and demographic patient factors on outcomes is unclear. We sought to identify socioeconomic and demographic patient characteristics associated with patient survival and tumor progression, and to characterize treatment options and healthcare utilization. Methods: A cohort of 169 patients with pathologically confirmed glioblastomas treated at our institution was retrospectively reviewed. Multivariable cox proportional hazards analysis for overall survival (OS) and cumulative incidence of progression was performed. Differences in treatment regimen, patient characteristics, and neuro-oncology office use between different age and depressive disorder history patient subgroups were calculated two-sample t-tests, Fisher's exact tests, or linear regression analysis. Results: The median age of all patients at the time of initiation of radiation therapy was 60.5 years. The median OS of the cohort was 13.1 months. Multivariable analysis identified age (Hazard Ratio 1.02, 95% CI 1.00-1.04) and total resection (Hazard Ratio 0.52, 95% CI 0.33-0.82) as significant predictors of OS. Increased number of radiation fractions (Hazard Ratio 0.90, 95% CI 0.82-0.98), depressive disorder history (Hazard Ratio 0.59, 95% CI 0.37-0.95), and total resection (Hazard Ratio 0.52, 95% CI 0.31-0.88) were associated with decreased incidence of progression. Notably, patients with depressive disorder history were observed to have more neuro-oncology physician office visits over time (median 12 vs. 16 visits, p = 0.0121). Patients older than 60 years and those with Medicare (vs. private) insurance were less likely to receive as many radiation fractions (p = 0.0014) or receive temozolomide concurrently with radiation (Odds Ratio 0.46, p = 0.0139). Conclusion: Older glioblastoma patients were less likely to receive as diverse of a treatment regimen as their younger counterparts, which may be partially driven by insurance type. Patients with depressive disorder history exhibited reduced incidence of progression, which may be due to more frequent health care contact during neuro-oncology physician office visits.

Neutrophilia and post-radiation thrombocytopenia predict for poor prognosis in radiation-treated glioma patients.

Introduction: Poor outcomes in glioma patients indicate a need to determine prognostic indicators of survival to better guide patient specific treatment options. While preoperative neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and monocyte-to-lymphocyte ratio (MLR) have been suggested as prognostic systemic inflammation markers, the impact of post-radiation changes in these cell types is unclear. We sought to identify which hematologic cell measurements before, during, or after radiation predicted for patient survival. Methods: A cohort of 182 patients with pathologically confirmed gliomas treated at our institution was retrospectively reviewed. Patient blood samples were collected within one month before, during, or within 3 months after radiation for quantification of hematologic cell counts, for which failure patterns were evaluated. Multivariable cox proportional hazards analysis for overall survival (OS) and progression-free survival (PFS) was performed to control for patient variables. Results: Multivariable analysis identified pre-radiation NLR > 4.0 (Hazard ratio = 1.847, p = 0.0039) and neutrophilia prior to (Hazard ratio = 1.706, p = 0.0185), during (Hazard ratio = 1.641, p = 0.0277), or after (Hazard ratio = 1.517, p = 0.0879) radiation as significant predictors of worse OS, with similar results for PFS. Post-radiation PLR > 200 (Hazard ratio = 0.587, p = 0.0062) and a percent increase in platelets after radiation (Hazard ratio = 0.387, p = 0.0077) were also associated with improved OS. Patients receiving more than 15 fractions of radiation exhibited greater post-radiation decreases in neutrophil and platelet counts than those receiving fewer. Patients receiving dexamethasone during radiation exhibited greater increases in neutrophil counts than those not receiving steroids. Lymphopenia, changes in lymphocyte counts, monocytosis, MLR, and changes in monocyte counts did not impact patient survival. Conclusion: Neutrophilia at any time interval surrounding radiotherapy, pre-radiation NLR, and post-radiation thrombocytopenia, but not lymphocytes or monocytes, are predictors of poor patient survival in glioma patients.

Chemical Exchange Saturation Transfer Magnetic Resonance Imaging Identifies Abnormal Calf Muscle-Specific Energetics in Peripheral Artery Disease.

BACKGROUND: Peripheral artery disease (PAD) results in exercise-induced ischemia in leg muscles. 31Phosphorus (P) magnetic resonance spectroscopy demonstrates prolonged phosphocreatine recovery time constant after exercise in PAD but has low signal to noise, low spatial resolution, and requires multinuclear hardware. Chemical exchange saturation transfer (CEST) is a quantitative magnetic resonance imaging method for imaging substrate (CEST asymmetry [CESTasym]) concentration by muscle group. We hypothesized that kinetics measured by CEST could distinguish between patients with PAD and controls. METHODS: Patients with PAD and age-matched normal subjects were imaged at 3T with a transmit-receive coil around the calf. Four CEST mages were acquired over 24-second intervals. The subjects then performed plantar flexion exercise on a magnetic resonance imaging-compatible ergometer until calf exhaustion. Twenty-five CEST images were obtained at end exercise. Regions of interest were drawn around individual muscle groups, and (CESTasym) decay times were fitted by exponential curve to CEST values. In 10 patients and 11 controls, 31P spectra were obtained 20 minutes later after repeat exercise. Five patients and 5 controls returned at a mean of 1±1 days later for repeat CEST studies. RESULTS: Thirty-five patients with PAD (31 male, age 66±8 years) and 29 controls (11 male, age 63±8 years) were imaged with CEST. The CESTasym decay times for the whole calf (341±332 versus 153±72 seconds; P<0.03) as well as for the gastrocnemius and posterior tibialis were longer in patients with PAD. Agreement between CESTasym decay and phosphocreatine recovery time constant was good. CONCLUSIONS: CEST is a magnetic resonance imaging method that can distinguish energetics in patients with PAD from age-matched normal subjects on a per muscle group basis. CEST agrees reasonably well with the gold standard 31P magnetic resonance spectroscopy. Moreover, CEST has higher spatial resolution, creates an image, and does not require multinuclear hardware and thus may be more suitable for clinical studies in PAD.

An interdisciplinary consensus on the management of brain metastases in patients with renal cell carcinoma.

Brain metastases are a challenging manifestation of renal cell carcinoma. We have a limited understanding of brain metastasis tumor and immune biology, drivers of resistance to systemic treatment, and their overall poor prognosis. Current data support a multimodal treatment strategy with radiation treatment and/or surgery. Nonetheless, the optimal approach for the management of brain metastases from renal cell carcinoma remains unclear. To improve patient care, the authors sought to standardize practical management strategies. They performed an unstructured literature review and elaborated on the current management strategies through an international group of experts from different disciplines assembled via the network of the International Kidney Cancer Coalition. Experts from different disciplines were administered a survey to answer questions related to current challenges and unmet patient needs. On the basis of the integrated approach of literature review and survey study results, the authors built algorithms for the management of single and multiple brain metastases in patients with renal cell carcinoma. The literature review, consensus statements, and algorithms presented in this report can serve as a framework guiding treatment decisions for patients. CA Cancer J Clin. 2022;72:454-489.

ZNF117 regulates glioblastoma stem cell differentiation towards oligodendroglial lineage.

Glioblastoma (GBM) is a deadly disease without effective treatment. Because glioblastoma stem cells (GSCs) contribute to tumor resistance and recurrence, improved treatment of GBM can be achieved by eliminating GSCs through inducing their differentiation. Prior efforts have been focused on studying GSC differentiation towards the astroglial lineage. However, regulation of GSC differentiation towards the neuronal and oligodendroglial lineages is largely unknown. To identify genes that control GSC differentiation to all three lineages, we performed an image-based genome-wide RNAi screen, in combination with single-cell RNA sequencing, and identified ZNF117 as a major regulator of GSC differentiation. Using patient-derived GSC cultures, we show that ZNF117 controls GSC differentiation towards the oligodendroglial lineage via the Notch pathway. We demonstrate that ZNF117 is a promising target for GSC differentiation therapy through targeted delivery of CRISPR/Cas9 gene-editing nanoparticles. Our study suggests a direction to improve GBM treatment through differentiation of GSCs towards various lineages.

Brain tumor IDH, 1p/19q, and MGMT molecular classification using MRI-based deep learning: an initial study on the effect of motion and motion correction.

Purpose: Deep learning has shown promise for predicting the molecular profiles of gliomas using MR images. Prior to clinical implementation, ensuring robustness to real-world problems, such as patient motion, is crucial. The purpose of this study is to perform a preliminary evaluation on the effects of simulated motion artifact on glioma marker classifier performance and determine if motion correction can restore classification accuracies. Approach: T2w images and molecular information were retrieved from the TCIA and TCGA databases. Simulated motion was added in the k-space domain along the phase encoding direction. Classifier performance for IDH mutation, 1p/19q co-deletion, and MGMT methylation was assessed over the range of 0% to 100% corrupted k-space lines. Rudimentary motion correction networks were trained on the motion-corrupted images. The performance of the three glioma marker classifiers was then evaluated on the motion-corrected images. Results: Glioma marker classifier performance decreased markedly with increasing motion corruption. Applying motion correction effectively restored classification accuracy for even the most motion-corrupted images. For isocitrate dehydrogenase (IDH) classification, 99% accuracy was achieved, exceeding the original performance of the network and representing a new benchmark in non-invasive MRI-based IDH classification. Conclusions: Robust motion correction can facilitate highly accurate deep learning MRI-based molecular marker classification, rivaling invasive tissue-based characterization methods. Motion correction may be able to increase classification accuracy even in the absence of a visible artifact, representing a new strategy for boosting classifier performance.

Right atrial volume index to left atrial volume index ratio is associated with adverse clinical outcomes in cardiogenic shock.

BACKGROUND: Structural remodeling in chronic systolic heart failure (HF) is associated with neurohormonal and hemodynamic perturbations among HF patients presenting with cardiogenic shock (CS) and HF. Our objective was to test the hypothesis was that atrial remodeling marked by an increased right atrial volume index (RAVI) to left atrial volume index (LAVI) ratio is associated with adverse clinical outcomes in CS. METHODS: Patients in this cohort were admitted to the intensive care unit with evidence of congestion (pulmonary capillary wedge pressure > 15) and cardiogenic shock (cardiac index < 2.2, systolic blood pressure < 90 mmHg, and clinical evidence supporting CS) and had an echocardiogram at the time of admission. RAVI was measured using Simpson's method in the apical four-chamber view, while LAVI was measured using the biplane disc summation method in the four and two-chamber views by two independent observers. Cox proportional hazards regression analysis was used to assess the association of RAVI-LAVI with the combined outcome of death or left ventricular assist device (LVAD). RESULTS: Among 113 patients (mean age 59 ± 14.9 years, 29.2% female), median RAVI/LAVI was 0.84. During a median follow-up of 12 months, 43 patients died, and 65 patients had the combined outcomes of death or LVAD. Patients with RAVI/LAVI ratio above the median had a greater incidence of death or LVAD (Log-rank p ≤ 0.001), and increasing RAVI/LAVI was significantly associated with the outcomes of death or LVAD (HR 1.71 95% CI 1.11-2.64, chi square 5.91, p = 0.010) even after adjustment for patient characteristics, echocardiographic and hemodynamic variables. CONCLUSION: RAVI/LAVI is an easily assessed novel echocardiographic parameter strongly associated with the survival and or the need for mechanical circulatory support in patients with CS.