Short- and Long-Term MRI Assessed Hemodynamic Changes in Pediatric Moyamoya Patients After Revascularization

BACKGROUND

Cerebrovascular reserve (CVR) reflects the capacity of cerebral blood flow (CBF) to change following a vasodilation challenge. Decreased CVR is associated with a higher stroke risk in patients with cerebrovascular diseases. While revascularization can improve CVR and reduce this risk in adult patients with vasculopathy such as those with Moyamoya disease, its impact on hemodynamics in pediatric patients remains to be elucidated. Arterial spin labeling (ASL) is a quantitative MRI technique that can measure CBF, CVR, and arterial transit time (ATT) non-invasively.

PURPOSE

To investigate the short- and long-term changes in hemodynamics after bypass surgeries in patients with Moyamoya disease.

STUDY TYPE

Longitudinal.

POPULATION

Forty-six patients (11 months-18 years, 28 females) with Moyamoya disease.

FIELD STRENGTH/SEQUENCE

3-T, single- and multi-delay ASL, T1-weighted, T2-FLAIR, 3D MRA.

ASSESSMENT

Imaging was performed 2 weeks before and 1 week and 6 months after surgical intervention. Acetazolamide was employed to induce vasodilation during the imaging procedure. CBF and ATT were measured by fitting the ASL data to the general kinetic model. CVR was computed as the percentage change in CBF. The mean CBF, ATT, and CVR values were measured in the regions affected by vasculopathy.

STATISTICAL TESTS

Pre- and post-revascularization CVR, CBF, and ATT were compared for different regions of the brain. P-values <0.05 were considered statistically significant.

RESULTS

ASL-derived CBF in flow territories affected by vasculopathy significantly increased after bypass by 41 ± 31% within a week. At 6 months, CBF significantly increased by 51 ± 34%, CVR increased by 68 ± 33%, and ATT was significantly reduced by 6.6 ± 2.9%.

DATA CONCLUSION

There may be short- and long-term improvement in the hemodynamic parameters of pediatric Moyamoya patients after bypass surgery.

EVIDENCE LEVEL

4 TECHNICAL EFFICACY: Stage 2.

Neuroimaging abnormalities associated with immunotherapy responsiveness in Down syndrome regression disorder

OBJECTIVE

To determine the prevalence of neuroimaging abnormalities in individuals with Down syndrome regression disorder (DSRD) and evaluate if neuroimaging abnormalities were predictive of therapeutic responses.

METHODS

A multicenter, retrospective, case-control study which reviewed neuroimaging studies of individuals with DSRD and compared them to a control cohort of individuals with Down syndrome (DS) alone was performed. Individuals aged 10-30 years and meeting international consensus criteria for DSRD were included. The presence of T1, T2/FLAIR, and SWI signal abnormalities was reviewed. Response rates to various therapies, including immunotherapy, were evaluated in the presence of neuroimaging abnormalities.

RESULTS

In total, 74 individuals (35%) had either T2/FLAIR and/or SWI signal abnormality compared to 14 individuals (12%) without DSRD (p < 0.001, 95%CI: 2.18-7.63). T2/FLAIR signal abnormalities were not appreciated more frequently in individuals with DSRD (14%, 30/210) than in the control cohort (9%, 11/119) (p = 0.18, OR: 1.63, 95%CI: 0.79-3.40). SWI signal abnormalities were appreciated at a higher frequency in individuals with DSRD (24%, 51/210) compared to the control cohort (4%, 5/119) (p < 0.001, OR: 7.31, 95%CI: 2.83-18.90). T2/FLAIR signal abnormalities were localized to the frontal (40%, 12/30) and parietal lobes (37%, 11/30). SWI signal abnormalities were predominantly in the bilateral basal ganglia (94%, 49/52). Individuals with DSRD and the presence of T2/FLAIR and/or SWI signal abnormalities were much more likely to respond to immunotherapy (p < 0.001, OR: 8.42. 95%CI: 3.78-18.76) and less likely to respond to benzodiazepines (p = 0.01, OR: 0.45, 95%CI: 0.25-0.83), antipsychotics (p < 0.001, OR: 0.28, 95%CI: 0.11-0.55), or electroconvulsive therapy (p < 0.001, OR: 0.12; 95%CI: 0.02-0.78) compared to individuals without these neuroimaging abnormalities.

INTERPRETATION

This study indicates that in individuals diagnosed with DSRD, T2/FLAIR, and SWI signal abnormalities are more common than previously thought and predict response to immunotherapy.

iSPAN: Explainable prediction of outcomes post thrombectomy with Machine Learning

PURPOSE

This study aimed to develop and evaluate a machine learning model and a novel clinical score for predicting outcomes in stroke patients undergoing endovascular thrombectomy.

MATERIALS AND METHODS

This retrospective study included all patients aged over 18 years with an anterior circulation stroke treated at a thrombectomy centre from 2010 to 2020 with external validation. The primary outcome was day 90 mRS ≥3. Existing clinical scores (SPAN and PRE) and Machine Learning (ML) models were compared. A novel clinical score (iSPAN) was derived by adding an optimised weighting of the most important ML features to the SPAN.

RESULTS

812 patients were initially included (397 female, average age 73), 63 for external validation. The best performing clinical score and ML model were SPAN and XGB (sensitivity, specificity and accuracy 0.290, 0.967, 0.628 and 0.693, 0.783, 0.738 respectively). A significant difference was found overall and our XGB model was more accurate than SPAN (p < 0.0018). The most important features were Age, mTICI and total number of passes. The addition of 11 points for mTICI of ≤2B and 3 points for ≥3 passes to the SPAN achieved the best accuracy and was used to create the iSPAN. iSPAN was not significantly less accurate than our XGB model (p > 0.5). In the external validation set, iSPAN and SPAN achieved sensitivity, specificity, and accuracy of (0.735, 0.862, 0.79) and (0.471, 0.897, 0.67) respectively.

CONCLUSION

iSPAN incorporates machine-derived features to achieve better predictions compared to existing clinical scores. It is not inferior to our XGB model and is externally generalisable.

Increased confidence of radiomics facilitating pretherapeutic differentiation of BRAF-altered pediatric low-grade glioma

OBJECTIVES

Currently, the BRAF status of pediatric low-grade glioma (pLGG) patients is determined through a biopsy. We established a nomogram to predict BRAF status non-invasively using clinical and radiomic factors. Additionally, we assessed an advanced thresholding method to provide only high-confidence predictions for the molecular subtype. Finally, we tested whether radiomic features provide additional predictive information for this classification task, beyond that which is embedded in the location of the tumor.

METHODS

Random forest (RF) models were trained on radiomic and clinical features both separately and together, to evaluate the utility of each feature set. Instead of using the traditional single threshold technique to convert the model outputs to class predictions, we implemented a double threshold mechanism that accounted for uncertainty. Additionally, a linear model was trained and depicted graphically as a nomogram.

RESULTS

The combined RF (AUC: 0.925) outperformed the RFs trained on radiomic (AUC: 0.863) or clinical (AUC: 0.889) features alone. The linear model had a comparable AUC (0.916), despite its lower complexity. Traditional thresholding produced an accuracy of 84.5%, while the double threshold approach yielded 92.2% accuracy on the 80.7% of patients with the highest confidence predictions.

CONCLUSION

Models that included radiomic features outperformed, underscoring their importance for the prediction of BRAF status. A linear model performed similarly to RF but with the added benefit that it can be visualized as a nomogram, improving the explainability of the model. The double threshold technique was able to identify uncertain predictions, enhancing the clinical utility of the model.

CLINICAL RELEVANCE STATEMENT

Radiomic features and tumor location are both predictive of BRAF status in pLGG patients. We show that they contain complementary information and depict the optimal model as a nomogram, which can be used as a non-invasive alternative to biopsy.

KEY POINTS

• Radiomic features provide additional predictive information for the determination of the molecular subtype of pediatric low-grade gliomas patients, beyond what is embedded in the location of the tumor, which has an established relationship with genetic status. • An advanced thresholding method can help to distinguish cases where machine learning models have a high chance of being (in)correct, improving the utility of these models. • A simple linear model performs similarly to a more powerful random forest model at classifying the molecular subtype of pediatric low-grade gliomas but has the added benefit that it can be converted into a nomogram, which may facilitate clinical implementation by improving the explainability of the model.

Microstructural Changes in the Brainstem Auditory Pathway in Children With Hearing Loss

OBJECTIVE

To assess the utility of diffusion tensor imaging of the auditory pathway in children with sensorineural hearing loss (SNHL).

STUDY DESIGN

Retrospective cohort study.

SETTING

A single academic tertiary children's hospital.

PATIENTS

Sixteen pediatric patients with bilateral SNHL of at least moderate severity in the poorer ear (eight male; mean age, 5.3 ± 4.9 yrs). Controls consisted of age- and sex-matched children with normal hearing who were imaged for nonotologic, non-neurologic medical concerns and found to have normal magnetic resonance imaging (MRI).

INTERVENTIONS

Three Tesla MRI scanners were used for diffusion tensor imaging.

MAIN OUTCOME MEASURES

Quantitative diffusion tensor metrics were extracted from the superior olivary nucleus (SON), inferior colliculus (IC), and ipsilateral fiber tracts between the SON and IC delineated by tractography.

RESULTS

We identified differences in fractional anisotropy of the SON between the SNHL cohort and controls (0.377 ± 0.056 vs. 0.422 ± 0.052; p = 0.009), but not in the IC. There were no differences in the mean diffusivity (MD) values in the IC and SON. Among younger children (≤5 yrs), MD was decreased in the SNHL cohort compared with controls in the IC (0.918 ± 0.051 vs. 1.120 ± 0.142; p < 0.001). However, among older children (>5 yrs), there were no differences in MD (1.124 ± 0.198 vs. 0.997 ± 0.103; p = 0.119). There were no differences in MD or fractional anisotropy in the white matter fibers of the IC-SON tract.

CONCLUSIONS

Our results suggest abnormal neural tracts along the central auditory pathway among children with SNHL. Longitudinal studies should assess the prognostic value of these MRI-based findings for assessing long-term outcomes and determining intervention efficacy.

A Radiomic "Warning Sign" of Progression on Brain MRI in Individuals with MS

BACKGROUND AND PURPOSE

MS is a chronic progressive, idiopathic, demyelinating disorder whose diagnosis is contingent on the interpretation of MR imaging. New MR imaging lesions are an early biomarker of disease progression. We aimed to evaluate a machine learning model based on radiomics features in predicting progression on MR imaging of the brain in individuals with MS.

MATERIALS AND METHODS

This retrospective cohort study with external validation on open-access data obtained full ethics approval. Longitudinal MR imaging data for patients with MS were collected and processed for machine learning. Radiomics features were extracted at the future location of a new lesion in the patients' prior MR imaging ("prelesion"). Additionally, "control" samples were obtained from the normal-appearing white matter for each participant. Machine learning models for binary classification were trained and tested and then evaluated the external data of the model.

RESULTS

The total number of participants was 167. Of the 147 in the training/test set, 102 were women and 45 were men. The average age was 42 (range, 21-74 years). The best-performing radiomics-based model was XGBoost, with accuracy, precision, recall, and F1-score of 0.91, 0.91, 0.91, and 0.91 on the test set, and 0.74, 0.74, 0.74, and 0.70 on the external validation set. The 5 most important radiomics features to the XGBoost model were associated with the overall heterogeneity and low gray-level emphasis of the segmented regions. Probability maps were produced to illustrate potential future clinical applications.

CONCLUSIONS

Our machine learning model based on radiomics features successfully differentiated prelesions from normal-appearing white matter. This outcome suggests that radiomics features from normal-appearing white matter could serve as an imaging biomarker for progression of MS on MR imaging.

Measuring Quantitative Cerebral Blood Flow in Healthy Children: A Systematic Review of Neuroimaging Techniques

Cerebral blood flow (CBF) is an important hemodynamic parameter to evaluate brain health. It can be obtained quantitatively using medical imaging modalities such as magnetic resonance imaging and positron emission tomography (PET). Although CBF in adults has been widely studied and linked with cerebrovascular and neurodegenerative diseases, CBF data in healthy children are sparse due to the challenges in pediatric neuroimaging. An understanding of the factors affecting pediatric CBF and its normal range is crucial to determine the optimal CBF measuring techniques in pediatric neuroradiology. This review focuses on pediatric CBF studies using neuroimaging techniques in 32 articles including 2668 normal subjects ranging from birth to 18 years old. A systematic literature search was conducted in PubMed, Embase, and Scopus and reported following the preferred reporting items for systematic reviews and meta-analyses (PRISMA). We identified factors (such as age, gender, mood, sedation, and fitness) that have significant effects on pediatric CBF quantification. We also investigated factors influencing the CBF measurements in infants. Based on this review, we recommend best practices to improve CBF measurements in pediatric neuroimaging. LEVEL OF EVIDENCE: 1 TECHNICAL EFFICACY: Stage 2.

Use of Convolutional Neural Networks to Evaluate Auricular Reconstruction Outcomes for Microtia

OBJECTIVES

The objective of this study is to determine whether machine learning may be used for objective assessment of aesthetic outcomes of auricular reconstructive surgery.

METHODS

Images of normal and reconstructed auricles were obtained from internet image search engines. Convolutional neural networks were constructed to identify auricles in 2D images in an auto-segmentation task and to evaluate whether an ear was normal versus reconstructed in a binary classification task. Images were then assigned a percent score for "normal" ear appearance based on confidence of the classification.

RESULTS

Images of 1115 ears (600 normal and 515 reconstructed) were obtained. The auto-segmentation task identified auricles with 95.30% accuracy compared to manually segmented auricles. The binary classification task achieved 89.22% accuracy in identifying reconstructed ears. When the confidence of the classification was used to assign percent scores to "normal" appearance, the reconstructed ears were classified to a range of 2% (least like normal ears) to 98% (most like normal ears).

CONCLUSION

Image-based analysis using machine learning can offer objective assessment without the bias of the patient or the surgeon. This methodology could be adapted to be used by surgeons to assess quality of operative outcome in clinical and research settings.

LEVEL OF EVIDENCE

4 Laryngoscope, 133:2413-2416, 2023.

The RSNA Cervical Spine Fracture CT Dataset

This dataset is composed of cervical spine CT images with annotations related to fractures; it is available at https://www.kaggle.com/competitions/rsna-2022-cervical-spine-fracture-detection/.

DEEP MOVEMENT: Deep learning of movie files for management of endovascular thrombectomy

OBJECTIVES

Treatment and outcomes of acute stroke have been revolutionised by mechanical thrombectomy. Deep learning has shown great promise in diagnostics but applications in video and interventional radiology lag behind. We aimed to develop a model that takes as input digital subtraction angiography (DSA) videos and classifies the video according to (1) the presence of large vessel occlusion (LVO), (2) the location of the occlusion, and (3) the efficacy of reperfusion.

METHODS

All patients who underwent DSA for anterior circulation acute ischaemic stroke between 2012 and 2019 were included. Consecutive normal studies were included to balance classes. An external validation (EV) dataset was collected from another institution. The trained model was also used on DSA videos post mechanical thrombectomy to assess thrombectomy efficacy.

RESULTS

In total, 1024 videos comprising 287 patients were included (44 for EV). Occlusion identification was achieved with 100% sensitivity and 91.67% specificity (EV 91.30% and 81.82%). Accuracy of location classification was 71% for ICA, 84% for M1, and 78% for M2 occlusions (EV 73, 25, and 50%). For post-thrombectomy DSA (n = 194), the model identified successful reperfusion with 100%, 88%, and 35% for ICA, M1, and M2 occlusion (EV 89, 88, and 60%). The model could also perform classification of post-intervention videos as mTICI < 3 with an AUC of 0.71.

CONCLUSIONS

Our model can successfully identify normal DSA studies from those with LVO and classify thrombectomy outcome and solve a clinical radiology problem with two temporal elements (dynamic video and pre and post intervention).

KEY POINTS

• DEEP MOVEMENT represents a novel application of a model applied to acute stroke imaging to handle two types of temporal complexity, dynamic video and pre and post intervention. • The model takes as an input digital subtraction angiograms of the anterior cerebral circulation and classifies according to (1) the presence or absence of large vessel occlusion, (2) the location of the occlusion, and (3) the efficacy of thrombectomy. • Potential clinical utility lies in providing decision support via rapid interpretation (pre thrombectomy) and automated objective gradation of thrombectomy outcomes (post thrombectomy).

Radiology artificial intelligence: a systematic review and evaluation of methods (RAISE)

OBJECTIVE

There has been a large amount of research in the field of artificial intelligence (AI) as applied to clinical radiology. However, these studies vary in design and quality and systematic reviews of the entire field are lacking.This systematic review aimed to identify all papers that used deep learning in radiology to survey the literature and to evaluate their methods. We aimed to identify the key questions being addressed in the literature and to identify the most effective methods employed.

METHODS

We followed the PRISMA guidelines and performed a systematic review of studies of AI in radiology published from 2015 to 2019. Our published protocol was prospectively registered.

RESULTS

Our search yielded 11,083 results. Seven hundred sixty-seven full texts were reviewed, and 535 articles were included. Ninety-eight percent were retrospective cohort studies. The median number of patients included was 460. Most studies involved MRI (37%). Neuroradiology was the most common subspecialty. Eighty-eight percent used supervised learning. The majority of studies undertook a segmentation task (39%). Performance comparison was with a state-of-the-art model in 37%. The most used established architecture was UNet (14%). The median performance for the most utilised evaluation metrics was Dice of 0.89 (range .49-.99), AUC of 0.903 (range 1.00-0.61) and Accuracy of 89.4 (range 70.2-100). Of the 77 studies that externally validated their results and allowed for direct comparison, performance on average decreased by 6% at external validation (range increase of 4% to decrease 44%).

CONCLUSION

This systematic review has surveyed the major advances in AI as applied to clinical radiology.

KEY POINTS

• While there are many papers reporting expert-level results by using deep learning in radiology, most apply only a narrow range of techniques to a narrow selection of use cases. • The literature is dominated by retrospective cohort studies with limited external validation with high potential for bias. • The recent advent of AI extensions to systematic reporting guidelines and prospective trial registration along with a focus on external validation and explanations show potential for translation of the hype surrounding AI from code to clinic.

Intracranial Hemorrhage in Term and Late-Preterm Neonates: An Institutional Perspective

BACKGROUND AND PURPOSE

Distribution of intracranial hemorrhage in term and late-preterm neonates is relatively unexplored. This descriptive study examines the MR imaging-detectable spectrum of intracranial hemorrhage in this population and potential risk factors.

MATERIALS AND METHODS

Prevalence and distribution of intracranial hemorrhage in consecutive term/late-preterm neonates who underwent brain MR imaging between January 2011 to August 2018 were assessed. MRIs were analyzed to determine intracranial hemorrhage distribution (intraventricular, subarachnoid, subdural, intraparenchymal, and subpial/leptomeningeal), and chart review was performed for potential clinical risk factors.

RESULTS

Of 725 term/late-preterm neonates who underwent brain MR imaging, intracranial hemorrhage occurred in 63 (9%). Fifty-two (83%) had multicompartment intracranial hemorrhage. Intraventricular and subdural were the most common hemorrhage locations, found in 41 (65%) and 39 (62%) neonates, respectively. Intraparenchymal hemorrhage occurred in 33 (52%); subpial, in 19 (30%); subarachnoid, in 12 (19%); and epidural, in 2 (3%) neonates. Twenty infants (32%) were delivered via cesarean delivery, and 5 (8%), via instrumented delivery. Cortical vein thromboses were present in 34 (54%); periventricular or medullary vein thromboses, in 37 (59%); and cerebral venous sinus thrombosis, in 5 (8%). Thirty-seven (59%) had elevated markers of coagulopathy (international normalized ratio > 1.2, fibrinogen level < 234), 9 (14%) had a clinically meaningful elevation in the international normalized ratio (>1.4), and 3 (5%) had a clinically meaningful decrease in the fibrinogen level (<150). Three (5%) neonates had thrombocytopenia (platelet count < 100 × 103/μL).

CONCLUSIONS

While relatively infrequent, there was a wide distribution of intracranial hemorrhage in term and late-preterm infants; intraventricular and subdural hemorrhages were the most common types. We report a high prevalence of venous congestion or thromboses accompanying neonatal intracranial hemorrhage.

A deep learning-based system for survival benefit prediction of tyrosine kinase inhibitors and immune checkpoint inhibitors in stage IV non-small cell lung cancer patients: A multicenter, prognostic study

BACKGROUND

For clinical decision making, it is crucial to identify patients with stage IV non-small cell lung cancer (NSCLC) who may benefit from tyrosine kinase inhibitors (TKIs) and immune checkpoint inhibitors (ICIs). In this study, a deep learning-based system was designed and validated using pre-therapy computed tomography (CT) images to predict the survival benefits of EGFR-TKIs and ICIs in stage IV NSCLC patients.

METHODS

This retrospective study collected data from 570 patients with stage IV EGFR-mutant NSCLC treated with EGFR-TKIs at five institutions between 2010 and 2021 (data of 314 patients were from a previously registered study), and 129 patients with stage IV NSCLC treated with ICIs at three institutions between 2017 and 2021 to build the ICI test dataset. Five-fold cross-validation was applied to divide the EGFR-TKI-treated patients from four institutions into training and internal validation datasets randomly in a ratio of 80%:20%, and the data from another institution was used as an external test dataset. An EfficientNetV2-based survival benefit prognosis (ESBP) system was developed with pre-therapy CT images as the input and the probability score as the output to identify which patients would receive additional survival benefit longer than the median PFS. Its prognostic performance was validated on the ICI test dataset. For diagnosing which patient would receive additional survival benefit, the accuracy of ESBP was compared with the estimations of three radiologists and three oncologists with varying degrees of expertise (two, five, and ten years). Improvements in the clinicians' diagnostic accuracy with ESBP assistance were then quantified.

FINDINGS

ESBP achieved positive predictive values of 80·40%, 75·40%, and 77·43% for additional EGFR-TKI survival benefit prediction using the probability score of 0·2 as the threshold on the training, internal validation, and external test datasets, respectively. The higher ESBP score (>0·2) indicated a better prognosis for progression-free survival (hazard ratio: 0·36, 95% CI: 0·19-0·68, p<0·0001) in patients on the external test dataset. Patients with scores >0·2 in the ICI test dataset also showed better survival benefit (hazard ratio: 0·33, 95% CI: 0·18-0·55, p<0·0001). This suggests the potential of ESBP to identify the two subgroups of benefiting patients by decoding the commonalities from pre-therapy CT images (stage IV EGFR-mutant NSCLC patients receiving additional survival benefit from EGFR-TKIs and stage IV NSCLC patients receiving additional survival benefit from ICIs). ESBP assistance improved the diagnostic accuracy of the clinicians with two years of experience from 47·91% to 66·32%, and the clinicians with five years of experience from 53·12% to 61·41%.

INTERPRETATION

This study developed and externally validated a preoperative CT image-based deep learning model to predict the survival benefits of EGFR-TKI and ICI therapies in stage IV NSCLC patients, which will facilitate optimized and individualized treatment strategies.

FUNDING

This study received funding from the National Natural Science Foundation of China (82001904, 81930053, and 62027901), and Key-Area Research and Development Program of Guangdong Province (2021B0101420005).

Intracranial Artery Morphology in Pediatric Moya Moya Disease and Moya Moya Syndrome

BACKGROUND

Moya Moya disease (MMD) and Moya Moya syndrome (MMS) are cerebrovascular disorders, which affect the internal carotid arteries (ICAs). Diagnosis and surveillance of MMD/MMS in children mostly rely on qualitative evaluation of vascular imaging, especially MR angiography (MRA).

OBJECTIVE

To quantitatively characterize arterial differences in pediatric patients with MMD/MMS compared with normal controls.

METHODS

MRA data sets from 17 presurgery MMD/MMS (10M/7F, mean age = 10.0 years) patients were retrospectively collected and compared with MRA data sets of 98 children with normal vessel morphology (49 male patients; mean age = 10.6 years). Using a level set segmentation method with anisotropic energy weights, the cerebral arteries were automatically extracted and used to compute the radius of the ICA, middle cerebral artery (MCA), anterior cerebral artery (ACA), posterior cerebral artery (PCA), and basilar artery (BA). Moreover, the density and the average radius of all arteries in the MCA, ACA, and PCA flow territories were quantified.

RESULTS

Statistical analysis revealed significant differences comparing children with MMD/MMS and those with normal vasculature (P < .001), whereas post hoc analyses identified significantly smaller radii of the ICA, MCA-M1, MCA-M2, and ACA (P < .001) in the MMD/MMS group. No significant differences were found for the radii of the PCA and BA or any artery density and average artery radius measurement in the flow territories (P > .05).

CONCLUSION

His study describes the results of an automatic approach for quantitative characterization of the cerebrovascular system in patients with MMD/MMS with promising preliminary results for quantitative surveillance in pediatric MMD/MMS management.

MRI Radiogenomics of Pediatric Medulloblastoma: A Multicenter Study

Background Radiogenomics of pediatric medulloblastoma (MB) offers an opportunity for MB risk stratification, which may aid therapeutic decision making, family counseling, and selection of patient groups suitable for targeted genetic analysis. Purpose To develop machine learning strategies that identify the four clinically significant MB molecular subgroups. Materials and Methods In this retrospective study, consecutive pediatric patients with newly diagnosed MB at MRI at 12 international pediatric sites between July 1997 and May 2020 were identified. There were 1800 features extracted from T2- and contrast-enhanced T1-weighted preoperative MRI scans. A two-stage sequential classifier was designed-one that first identifies non-wingless (WNT) and non-sonic hedgehog (SHH) MB and then differentiates therapeutically relevant WNT from SHH. Further, a classifier that distinguishes high-risk group 3 from group 4 MB was developed. An independent, binary subgroup analysis was conducted to uncover radiomics features unique to infantile versus childhood SHH subgroups. The best-performing models from six candidate classifiers were selected, and performance was measured on holdout test sets. CIs were obtained by bootstrapping the test sets for 2000 random samples. Model accuracy score was compared with the no-information rate using the Wald test. Results The study cohort comprised 263 patients (mean age ± SD at diagnosis, 87 months ± 60; 166 boys). A two-stage classifier outperformed a single-stage multiclass classifier. The combined, sequential classifier achieved a microaveraged F1 score of 88% and a binary F1 score of 95% specifically for WNT. A group 3 versus group 4 classifier achieved an area under the receiver operating characteristic curve of 98%. Of the Image Biomarker Standardization Initiative features, texture and first-order intensity features were most contributory across the molecular subgroups. Conclusion An MRI-based machine learning decision path allowed identification of the four clinically relevant molecular pediatric medulloblastoma subgroups. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Chaudhary and Bapuraj in this issue.

Early Onset Diffusion Abnormalities in Refractory Headache Disorders

Objective

This study sought to determine if individuals with medically refractory migraine headache have volume or diffusion abnormalities on neuroimaging compared to neurotypical individuals.

Background

Neuroimaging biomarkers in headache medicine continue to be limited. Early prediction of medically refractory headache and migraine disorders could result in earlier administration of high efficacy therapeutics.

Methods

A single-center, retrospective, case control study was performed. All patients were evaluated clinically between 2014 and 2018. Individuals with medically refractory migraine headache (defined by ICDH-3 criteria) without any other chronic medical diseases were enrolled. Patients had to have failed more than two therapeutics and aura was not exclusionary. The initial MRI study for each patient was reviewed. Multiple brain regions were analyzed for volume and apparent diffusion coefficient values. These were compared to 81 neurotypical control patients.

Results

A total of 79 patients with medically refractory migraine headache were included and compared to 74 neurotypical controls without headache disorders. Time between clinical diagnosis and neuroimaging was a median of 24 months (IQR: 12.0–37.0). Comparison of individuals with medically refractory migraine headache to controls revealed statistically significant differences in median apparent diffusion coefficient (ADC) in multiple brain subregions (p < 0.001). Post-hoc pair-wise analysis comparing individuals with medically refractory migraine headache to control patients revealed significantly decreased median ADC values for the thalamus, caudate, putamen, pallidum, amygdala, brainstem, and cerebral white matter. No volumetric differences were observed between groups.

Conclusions

In individuals with medically refractory MH, ADC changes are measurable in multiple brain structures at an early age, prior to the failure of multiple pharmacologic interventions and the diagnosis of medically refractory MH. This data supports the hypothesis that structural connectivity issues may predispose some patients toward more medically refractory pain disorders such as MH.

Radiomics Can Distinguish Pediatric Supratentorial Embryonal Tumors, High-Grade Gliomas, and Ependymomas

BACKGROUND AND PURPOSE

Pediatric supratentorial tumors such as embryonal tumors, high-grade gliomas, and ependymomas are difficult to distinguish by histopathology and imaging because of overlapping features. We applied machine learning to uncover MR imaging-based radiomics phenotypes that can differentiate these tumor types.

MATERIALS AND METHODS

Our retrospective cohort of 231 patients from 7 participating institutions had 50 embryonal tumors, 127 high-grade gliomas, and 54 ependymomas. For each tumor volume, we extracted 900 Image Biomarker Standardization Initiative-based PyRadiomics features from T2-weighted and gadolinium-enhanced T1-weighted images. A reduced feature set was obtained by sparse regression analysis and was used as input for 6 candidate classifier models. Training and test sets were randomly allocated from the total cohort in a 75:25 ratio.

RESULTS

The final classifier model for embryonal tumor-versus-high-grade gliomas identified 23 features with an area under the curve of 0.98; the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.85, 0.91, 0.79, 0.94, and 0.89, respectively. The classifier for embryonal tumor-versus-ependymomas identified 4 features with an area under the curve of 0.82; the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.93, 0.69, 0.76, 0.90, and 0.81, respectively. The classifier for high-grade gliomas-versus-ependymomas identified 35 features with an area under the curve of 0.96; the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 0.82, 0.94, 0.82, 0.94, and 0.91, respectively.

CONCLUSIONS

In this multi-institutional study, we identified distinct radiomic phenotypes that distinguish pediatric supratentorial tumors, high-grade gliomas, and ependymomas with high accuracy. Incorporation of this technique in diagnostic algorithms can improve diagnosis, risk stratification, and treatment planning.

Spatiotemporal changes in along-tract profilometry of cerebellar peduncles in cerebellar mutism syndrome

Cerebellar mutism syndrome, characterised by mutism, emotional lability and cerebellar motor signs, occurs in up to 39% of children following resection of medulloblastoma, the most common malignant posterior fossa tumour of childhood. Its pathophysiology remains unclear, but prior studies have implicated damage to the superior cerebellar peduncles. In this study, the objective was to conduct high-resolution spatial profilometry of the cerebellar peduncles and identify anatomic biomarkers of cerebellar mutism syndrome. In this retrospective study, twenty-eight children with medulloblastoma (mean age 8.8 ± 3.8 years) underwent diffusion MRI at four timepoints over one year. Forty-nine healthy children (9.0 ± 4.2 years), scanned at a single timepoint, served as age- and sex-matched controls. Automated Fibre Quantification was used to segment cerebellar peduncles and compute fractional anisotropy (FA) at 30 nodes along each tract. Thirteen patients developed cerebellar mutism syndrome. FA was significantly lower in the distal third of the left superior cerebellar peduncle pre-operatively in all patients compared to controls (FA in proximal third 0.228, middle and distal thirds 0.270, p = 0.01, Cohen's d = 0.927). Pre-operative differences in FA did not predict cerebellar mutism syndrome. However, post-operative reductions in FA were highly specific to the distal left superior cerebellar peduncle, and were most pronounced in children with cerebellar mutism syndrome compared to those without at the 1-4 month follow up (0.325 vs 0.512, p = 0.042, d = 1.36) and at the 1-year follow up (0.342, vs 0.484, p = 0.038, d = 1.12). High spatial resolution cerebellar profilometry indicated a site-specific alteration of the distal segment of the superior cerebellar peduncle seen in cerebellar mutism syndrome which may have important surgical implications in the treatment of these devastating tumours of childhood.

GD2-CAR T cell therapy for H3K27M-mutated diffuse midline gliomas

Diffuse intrinsic pontine glioma (DIPG) and other H3K27M-mutated diffuse midline gliomas (DMGs) are universally lethal paediatric tumours of the central nervous system¹. We have previously shown that the disialoganglioside GD2 is highly expressed on H3K27M-mutated glioma cells and have demonstrated promising preclinical efficacy of GD2-directed chimeric antigen receptor (CAR) T cells², providing the rationale for a first-in-human phase I clinical trial (NCT04196413). Because CAR T cell-induced brainstem inflammation can result in obstructive hydrocephalus, increased intracranial pressure and dangerous tissue shifts, neurocritical care precautions were incorporated. Here we present the clinical experience from the first four patients with H3K27M-mutated DIPG or spinal cord DMG treated with GD2-CAR T cells at dose level 1 (1 × 10⁶ GD2-CAR T cells per kg administered intravenously). Patients who exhibited clinical benefit were eligible for subsequent GD2-CAR T cell infusions administered intracerebroventricularly³. Toxicity was largely related to the location of the tumour and was reversible with intensive supportive care. On-target, off-tumour toxicity was not observed. Three of four patients exhibited clinical and radiographic improvement. Pro-inflammatory cytokine levels were increased in the plasma and cerebrospinal fluid. Transcriptomic analyses of 65,598 single cells from CAR T cell products and cerebrospinal fluid elucidate heterogeneity in response between participants and administration routes. These early results underscore the promise of this therapeutic approach for patients with H3K27M-mutated DIPG or spinal cord DMG.

MRI Correlates of Ototoxicity in the Auditory Pathway in Children Treated for Medulloblastoma

OBJECTIVE

To assess diffusion and perfusion changes of the auditory pathway in pediatric medulloblastoma patients exposed to ototoxic therapies.

STUDY DESIGN

Retrospective cohort study.

SETTING

A single academic tertiary children's hospital.

PATIENTS

Twenty pediatric medulloblastoma patients (13 men; mean age 12.0 ± 4.8 yr) treated with platinum-based chemotherapy with or without radiation and 18 age-and-sex matched controls were included. Ototoxicity scores were determined using Chang Ototoxicity Grading Scale.

INTERVENTIONS

Three Tesla magnetic resonance was used for diffusion tensor and arterial spin labeling perfusion imaging.

MAIN OUTCOME MEASURES

Quantitative diffusion tensor metrics were extracted from the Heschl's gyrus, auditory radiation, and inferior colliculus. Arterial spin labeling perfusion of the Heschl's gyrus was also examined.

RESULTS

Nine patients had clinically significant hearing loss, or Chang grades more than or equal to 2a; 11 patients had mild/no hearing loss, or Chang grades less than 2a. The clinically significant hearing loss group showed reduced mean diffusivity in the Heschl's gyrus (p = 0.018) and auditory radiation (p = 0.037), and decreased perfusion in the Heschl's gyrus (p = 0.001). Mild/no hearing loss group showed reduced mean diffusivity (p = 0.036) in Heschl's gyrus only, with a decrease in perfusion (p = 0.008). There were no differences between groups in the inferior colliculus. There was no difference in fractional anisotropy between patients exposed to ototoxic therapies and controls.

CONCLUSIONS

Patients exposed to ototoxic therapies demonstrated microstructural and physiological alteration of the auditory pathway. The present study shows proof-of-concept use of diffusion tensor imaging to gauge ototoxicity along the auditory pathway. Future larger cohort studies are needed to assess significance of changes in diffusion tensor imaging longitudinally, and the relationship between these changes and hearing loss severity and longitudinal changes of the developing auditory white matter.

Automatic lung nodule segmentation and intra-nodular heterogeneity image generation

Automatic segmentation of lung nodules on computed tomography (CT) images is challenging owing to the variability of morphology, location, and intensity. In addition, few segmentation methods can capture intra-nodular heterogeneity to assist lung nodule diagnosis. In this study, we propose an end-to-end architecture to perform fully automated segmentation of multiple types of lung nodules and generate intra-nodular heterogeneity images for clinical use. To this end, a hybrid loss is considered by introducing a Faster R-CNN model based on generalized intersection over union loss in generative adversarial network. The Lung Image Database Consortium image collection dataset, comprising 2,635 lung nodules, was combined with 3,200 lung nodules from five hospitals for this study. Compared with manual segmentation by radiologists, the proposed model obtained an average dice coefficient (DC) of 82.05% on the test dataset. Compared with U-net, NoduleNet, nnU-net, and other three models, the proposed method achieved comparable performance on lung nodule segmentation and generated more vivid and valid intra-nodular heterogeneity images, which are beneficial in radiological diagnosis. In an external test of 91 patients from another hospital, the proposed model achieved an average DC of 81.61%. The proposed method effectively addresses the challenges of inevitable human interaction and additional pre-processing procedures in the existing solutions for lung nodule segmentation. In addition, the results show that the intra-nodular heterogeneity images generated by the proposed model are suitable to facilitate lung nodule diagnosis in radiology.

Acetazolamide-Challenged Arterial Spin Labeling Detects Augmented Cerebrovascular Reserve After Surgery for Moyamoya

BACKGROUND AND PURPOSE

Cerebrovascular reserve (CVR) inversely correlates with stroke risk in children with Moyamoya disease and may be improved by revascularization surgery. We hypothesized that acetazolamide-challenged arterial spin labeling MR perfusion quantifies augmentation of CVR achieved by revascularization and correlates with currently accepted angiographic scoring criteria.

METHODS

We retrospectively identified pediatric patients with Moyamoya disease or syndrome who received cerebral revascularization at ≤18 years of age between 2012 and 2019 at our institution. Using acetazolamide-challenged arterial spin labeling, we compared postoperative CVR to corresponding preoperative values and to postoperative perfusion outcomes classified by Matsushima grading.

RESULTS

In this cohort, 32 patients (17 males) with Moyamoya underwent 29 direct and 16 indirect extracranial-intracranial bypasses at a median 9.7 years of age (interquartile range, 7.6-15.7). Following revascularization, median CVR increased within the ipsilateral middle cerebral artery territory (6.9 mL/100 g per minute preoperatively versus 16.5 mL/100 g per minute postoperatively, P<0.01). No differences were observed in the ipsilateral anterior cerebral artery (P=0.13) and posterior cerebral artery (P=0.48) territories. Postoperative CVR was higher in the ipsilateral middle cerebral artery territories of patients who achieved Matsushima grade A perfusion, in comparison to those with grades B or C (25.8 versus 17.5 mL, P=0.02). The method of bypass (direct or indirect) did not alter relative increases in CVR (8 versus 3.8 mL/100 g per minute, P=0.7).

CONCLUSIONS

Acetazolamide-challenged arterial spin labeling noninvasively quantifies augmentation of CVR following surgery for Moyamoya disease and syndrome.

Radiomic Signatures of Posterior Fossa Ependymoma: Molecular Subgroups and Risk Profiles

BACKGROUND

The risk profile for posterior fossa ependymoma (EP) depends on surgical and molecular status [Group A (PFA) versus Group B (PFB)]. While subtotal tumor resection is known to confer worse prognosis, MRI-based EP risk-profiling is unexplored. We aimed to apply machine learning strategies to link MRI-based biomarkers of high-risk EP and also to distinguish PFA from PFB.

METHODS

We extracted 1800 quantitative features from presurgical T2-weighted (T2-MRI) and gadolinium-enhanced T1-weighted (T1-MRI) imaging of 157 EP patients. We implemented nested cross-validation to identify features for risk score calculations and apply a Cox model for survival analysis. We conducted additional feature selection for PFA versus PFB and examined performance across three candidate classifiers.

RESULTS

For all EP patients with GTR, we identified four T2-MRI-based features and stratified patients into high- and low-risk groups, with 5-year overall survival rates of 62% and 100%, respectively (p < 0.0001). Among presumed PFA patients with GTR, four T1-MRI and five T2-MRI features predicted divergence of high- and low-risk groups, with 5-year overall survival rates of 62.7% and 96.7%, respectively (p = 0.002). T1-MRI-based features showed the best performance distinguishing PFA from PFB with an AUC of 0.86.

CONCLUSIONS

We present machine learning strategies to identify MRI phenotypes that distinguish PFA from PFB, as well as high- and low-risk PFA. We also describe quantitative image predictors of aggressive EP tumors that might assist risk-profiling after surgery. Future studies could examine translating radiomics as an adjunct to EP risk assessment when considering therapy strategies or trial candidacy.

Clinical Artificial Intelligence Applications in Radiology: Neuro

Radiologists have been at the forefront of the digitization process in medicine. Artificial intelligence (AI) is a promising area of innovation, particularly in medical imaging. The number of applications of AI in neuroradiology has also grown. This article illustrates some of these applications. This article reviews machine learning challenges related to neuroradiology. The first approval of reimbursement for an AI algorithm by the Centers for Medicare and Medicaid Services, covering a stroke software for early detection of large vessel occlusion, is also discussed.

Machine Learning Approach to Differentiation of Peripheral Schwannomas and Neurofibromas: A Multi-Center Study

BACKGROUND

Non-invasive differentiation between schwannomas and neurofibromas is important for appropriate management, preoperative counseling, and surgical planning, but has proven difficult using conventional imaging. The objective of this study was to develop and evaluate machine learning approaches for differentiating peripheral schwannomas from neurofibromas.

METHODS

We assembled a cohort of schwannomas and neurofibromas from 3 independent institutions and extracted high-dimensional radiomic features from gadolinium-enhanced, T1-weighted MRI using the PyRadiomics package on Quantitative Imaging Feature Pipeline. Age, sex, neurogenetic syndrome, spontaneous pain, and motor deficit were recorded. We evaluated the performance of 6 radiomics-based classifier models with and without clinical features and compared model performance against human expert evaluators.

RESULTS

107 schwannomas and 59 neurofibroma were included. The primary models included both clinical and imaging data. The accuracy of the human evaluators (0.765) did not significantly exceed the no-information rate (NIR), whereas the Support Vector Machine (0.929), Logistic Regression (0.929), and Random Forest (0.905) classifiers exceeded the NIR. Using the method of DeLong, the AUC for the Logistic Regression (AUC=0.923) and K Nearest Neighbor (AUC=0.923) classifiers was significantly greater than the human evaluators (AUC=0.766; p = 0.041).

CONCLUSIONS

The radiomics-based classifiers developed here proved to be more accurate and had a higher AUC on the ROC curve than expert human evaluators. This demonstrates that radiomics using routine MRI sequences and clinical features can aid in differentiation of peripheral schwannomas and neurofibromas.

Effects of Age on White Matter Microstructure in Children With Neurofibromatosis Type 1

Children with neurofibromatosis type 1 (NF1) often report cognitive challenges, though the etiology of such remains an area of active investigation. With the advent of treatments that may affect white matter microstructure, understanding the effects of age on white matter aberrancies in NF1 becomes crucial in determining the timing of such therapeutic interventions. A cross-sectional study was performed with diffusion tensor imaging from 18 NF1 children and 26 age-matched controls. Fractional anisotropy was determined by region of interest analyses for both groups over the corpus callosum, cingulate, and bilateral frontal and temporal white matter regions. Two-way analyses of variance were done with both ages combined and age-stratified into early childhood, middle childhood, and adolescence. Significant differences in fractional anisotropy between NF1 and controls were seen in the corpus callosum and frontal white matter regions when ages were combined. When stratified by age, we found that this difference was largely driven by the early childhood (1-5.9 years) and middle childhood (6-11.9 years) age groups, whereas no significant differences were appreciable in the adolescence age group (12-18 years). This study demonstrates age-related effects on white matter microstructure disorganization in NF1, suggesting that the appropriate timing of therapeutic intervention may be in early childhood.

Machine-Learning Approach to Differentiation of Benign and Malignant Peripheral Nerve Sheath Tumors: A Multicenter Study

BACKGROUND

Clinicoradiologic differentiation between benign and malignant peripheral nerve sheath tumors (PNSTs) has important management implications.

OBJECTIVE

To develop and evaluate machine-learning approaches to differentiate benign from malignant PNSTs.

METHODS

We identified PNSTs treated at 3 institutions and extracted high-dimensional radiomics features from gadolinium-enhanced, T1-weighted magnetic resonance imaging (MRI) sequences. Training and test sets were selected randomly in a 70:30 ratio. A total of 900 image features were automatically extracted using the PyRadiomics package from Quantitative Imaging Feature Pipeline. Clinical data including age, sex, neurogenetic syndrome presence, spontaneous pain, and motor deficit were also incorporated. Features were selected using sparse regression analysis and retained features were further refined by gradient boost modeling to optimize the area under the curve (AUC) for diagnosis. We evaluated the performance of radiomics-based classifiers with and without clinical features and compared performance against human readers.

RESULTS

A total of 95 malignant and 171 benign PNSTs were included. The final classifier model included 21 imaging and clinical features. Sensitivity, specificity, and AUC of 0.676, 0.882, and 0.845, respectively, were achieved on the test set. Using imaging and clinical features, human experts collectively achieved sensitivity, specificity, and AUC of 0.786, 0.431, and 0.624, respectively. The AUC of the classifier was statistically better than expert humans (P = .002). Expert humans were not statistically better than the no-information rate, whereas the classifier was (P = .001).

CONCLUSION

Radiomics-based machine learning using routine MRI sequences and clinical features can aid in evaluation of PNSTs. Further improvement may be achieved by incorporating additional imaging sequences and clinical variables into future models.

Machine Assist for Pediatric Posterior Fossa Tumor Diagnosis: A Multinational Study

BACKGROUND

Clinicians and machine classifiers reliably diagnose pilocytic astrocytoma (PA) on magnetic resonance imaging (MRI) but less accurately distinguish medulloblastoma (MB) from ependymoma (EP). One strategy is to first rule out the most identifiable diagnosis.

OBJECTIVE

To hypothesize a sequential machine-learning classifier could improve diagnostic performance by mimicking a clinician's strategy of excluding PA before distinguishing MB from EP.

METHODS

We extracted 1800 total Image Biomarker Standardization Initiative (IBSI)-based features from T2- and gadolinium-enhanced T1-weighted images in a multinational cohort of 274 MB, 156 PA, and 97 EP. We designed a 2-step sequential classifier - first ruling out PA, and next distinguishing MB from EP. For each step, we selected the best performing model from 6-candidate classifier using a reduced feature set, and measured performance on a holdout test set with the microaveraged F1 score.

RESULTS

Optimal diagnostic performance was achieved using 2 decision steps, each with its own distinct imaging features and classifier method. A 3-way logistic regression classifier first distinguished PA from non-PA, with T2 uniformity and T1 contrast as the most relevant IBSI features (F1 score 0.8809). A 2-way neural net classifier next distinguished MB from EP, with T2 sphericity and T1 flatness as most relevant (F1 score 0.9189). The combined, sequential classifier was with F1 score 0.9179.

CONCLUSION

An MRI-based sequential machine-learning classifiers offer high-performance prediction of pediatric posterior fossa tumors across a large, multinational cohort. Optimization of this model with demographic, clinical, imaging, and molecular predictors could provide significant advantages for family counseling and surgical planning.

Radiomic Phenotypes Distinguish Atypical Teratoid/Rhabdoid Tumors from Medulloblastoma

BACKGROUND AND PURPOSE

Atypical teratoid/rhabdoid tumors and medulloblastomas have similar imaging and histologic features but distinctly different outcomes. We hypothesized that they could be distinguished by MR imaging-based radiomic phenotypes.

MATERIALS AND METHODS

We retrospectively assembled T2-weighted and gadolinium-enhanced T1-weighted images of 48 posterior fossa atypical teratoid/rhabdoid tumors and 96 match-paired medulloblastomas from 7 institutions. Using a holdout test set, we measured the performance of 6 candidate classifier models using 6 imaging features derived by sparse regression of 900 T2WI and 900 T1WI Imaging Biomarker Standardization Initiative-based radiomics features.

RESULTS

From the originally extracted 1800 total Imaging Biomarker Standardization Initiative-based features, sparse regression consistently reduced the feature set to 1 from T1WI and 5 from T2WI. Among classifier models, logistic regression performed with the highest AUC of 0.86, with sensitivity, specificity, accuracy, and F1 scores of 0.80, 0.82, 0.81, and 0.85, respectively. The top 3 important Imaging Biomarker Standardization Initiative features, by decreasing order of relative contribution, included voxel intensity at the 90th percentile, inverse difference moment normalized, and kurtosis-all from T2WI.

CONCLUSIONS

Six quantitative signatures of image intensity, texture, and morphology distinguish atypical teratoid/rhabdoid tumors from medulloblastomas with high prediction performance across different machine learning strategies. Use of this technique for preoperative diagnosis of atypical teratoid/rhabdoid tumors could significantly inform therapeutic strategies and patient care discussions.

MRI-based radiomics for prognosis of pediatric diffuse intrinsic pontine glioma: an international study

Background

Diffuse intrinsic pontine gliomas (DIPGs) are lethal pediatric brain tumors. Presently, MRI is the mainstay of disease diagnosis and surveillance. We identify clinically significant computational features from MRI and create a prognostic machine learning model.

Methods

We isolated tumor volumes of T1-post-contrast (T1) and T2-weighted (T2) MRIs from 177 treatment-naïve DIPG patients from an international cohort for model training and testing. The Quantitative Image Feature Pipeline and PyRadiomics was used for feature extraction. Ten-fold cross-validation of least absolute shrinkage and selection operator Cox regression selected optimal features to predict overall survival in the training dataset and tested in the independent testing dataset. We analyzed model performance using clinical variables (age at diagnosis and sex) only, radiomics only, and radiomics plus clinical variables.

Results

All selected features were intensity and texture-based on the wavelet-filtered images (3 T1 gray-level co-occurrence matrix (GLCM) texture features, T2 GLCM texture feature, and T2 first-order mean). This multivariable Cox model demonstrated a concordance of 0.68 (95% CI: 0.61–0.74) in the training dataset, significantly outperforming the clinical-only model (C = 0.57 [95% CI: 0.49–0.64]). Adding clinical features to radiomics slightly improved performance (C = 0.70 [95% CI: 0.64–0.77]). The combined radiomics and clinical model was validated in the independent testing dataset (C = 0.59 [95% CI: 0.51–0.67], Noether’s test P = .02).

Conclusions

In this international study, we demonstrate the use of radiomic signatures to create a machine learning model for DIPG prognostication. Standardized, quantitative approaches that objectively measure DIPG changes, including computational MRI evaluation, could offer new approaches to assessing tumor phenotype and serve a future role for optimizing clinical trial eligibility and tumor surveillance.

Multi-classifier-based identification of COVID-19 from chest computed tomography using generalizable and interpretable radiomics features

PURPOSE

To investigate the efficacy of radiomics in diagnosing patients with coronavirus disease (COVID-19) and other types of viral pneumonia with clinical symptoms and CT signs similar to those of COVID-19.

METHODS

Between 18 January 2020 and 20 May 2020, 110 SARS-CoV-2 positive and 108 SARS-CoV-2 negative patients were retrospectively recruited from three hospitals based on the inclusion criteria. Manual segmentation of pneumonia lesions on CT scans was performed by four radiologists. The latest version of Pyradiomics was used for feature extraction. Four classifiers (linear classifier, k-nearest neighbour, least absolute shrinkage and selection operator [LASSO], and random forest) were used to differentiate SARS-CoV-2 positive and SARS-CoV-2 negative patients. Comparison of the performance of the classifiers and radiologists was evaluated by ROC curve and Kappa score.

RESULTS

We manually segmented 16,053 CT slices, comprising 32,625 pneumonia lesions, from the CT scans of all patients. Using Pyradiomics, 120 radiomic features were extracted from each image. The key radiomic features screened by different classifiers varied and lead to significant differences in classification accuracy. The LASSO achieved the best performance (sensitivity: 72.2%, specificity: 75.1%, and AUC: 0.81) on the external validation dataset and attained excellent agreement (Kappa score: 0.89) with radiologists (average sensitivity: 75.6%, specificity: 78.2%, and AUC: 0.81). All classifiers indicated that "Original_Firstorder_RootMeanSquared" and "Original_Firstorder_Uniformity" were significant features for this task.

CONCLUSIONS

We identified radiomic features that were significantly associated with the classification of COVID-19 pneumonia using multiple classifiers. The quantifiable interpretation of the differences in features between the two groups extends our understanding of CT imaging characteristics of COVID-19 pneumonia.

Radiomics of Pediatric Low-Grade Gliomas: Toward a Pretherapeutic Differentiation of BRAF-Mutated and BRAF-Fused Tumors

BACKGROUND AND PURPOSE

B-Raf proto-oncogene, serine/threonine kinase (BRAF) status has important implications for prognosis and therapy of pediatric low-grade gliomas. Currently, BRAF status classification relies on biopsy. Our aim was to train and validate a radiomics approach to predict BRAF fusion and BRAF V600E mutation.

MATERIALS AND METHODS

In this bi-institutional retrospective study, FLAIR MR imaging datasets of 115 pediatric patients with low-grade gliomas from 2 children's hospitals acquired between January 2009 and January 2016 were included and analyzed. Radiomics features were extracted from tumor segmentations, and the predictive model was tested using independent training and testing datasets, with all available tumor types. The model was selected on the basis of a grid search on the number of trees, opting for the best split for a random forest. We used the area under the receiver operating characteristic curve to evaluate model performance.

RESULTS

The training cohort consisted of 94 pediatric patients with low-grade gliomas (mean age, 9.4 years; 45 boys), and the external validation cohort comprised 21 pediatric patients with low-grade gliomas (mean age, 8.37 years; 12 boys). A 4-fold cross-validation scheme predicted BRAF status with an area under the curve of 0.75 (SD, 0.12) (95% confidence interval, 0.62-0.89) on the internal validation cohort. By means of the optimal hyperparameters determined by 4-fold cross-validation, the area under the curve for the external validation was 0.85. Age and tumor location were significant predictors of BRAF status (P values = .04 and <.001, respectively). Sex was not a significant predictor (P value = .96).

CONCLUSIONS

Radiomics-based prediction of BRAF status in pediatric low-grade gliomas appears feasible in this bi-institutional exploratory study.

Molecular correlates of cerebellar mutism syndrome in medulloblastoma

BACKGROUND

Cerebellar mutism syndrome (CMS) is a common complication following resection of posterior fossa tumors, most commonly after surgery for medulloblastoma. Medulloblastoma subgroups have historically been treated as a single entity when assessing CMS risk; however, recent studies highlighting their clinical heterogeneity suggest the need for subgroup-specific analysis. Here, we examine a large international multicenter cohort of molecularly characterized medulloblastoma patients to assess predictors of CMS.

METHODS

We assembled a cohort of 370 molecularly characterized medulloblastoma subjects with available neuroimaging from 5 sites globally, including Great Ormond Street Hospital, Christian Medical College and Hospital, the Hospital for Sick Children, King Hussein Cancer Center, and Lucile Packard Children's Hospital. Age at diagnosis, sex, tumor volume, and CMS development were assessed in addition to molecular subgroup.

RESULTS

Overall, 23.8% of patients developed CMS. CMS patients were younger (mean difference -2.05 years ± 0.50, P = 0.0218) and had larger tumors (mean difference 10.25 cm3 ± 4.60, P = 0.0010) that were more often midline (odds ratio [OR] = 5.72, P < 0.0001). In a multivariable analysis adjusting for age, sex, midline location, and tumor volume, Wingless (adjusted OR = 4.91, P = 0.0063), Group 3 (adjusted OR = 5.56, P = 0.0022), and Group 4 (adjusted OR = 8.57 P = 9.1 × 10-5) tumors were found to be independently associated with higher risk of CMS compared with sonic hedgehog tumors.

CONCLUSIONS

Medulloblastoma subgroup is a very strong predictor of CMS development, independent of tumor volume and midline location. These findings have significant implications for management of both the tumor and CMS.

Deep COVID DeteCT: an international experience on COVID-19 lung detection and prognosis using chest CT

The Coronavirus disease 2019 (COVID-19) presents open questions in how we clinically diagnose and assess disease course. Recently, chest computed tomography (CT) has shown utility for COVID-19 diagnosis. In this study, we developed Deep COVID DeteCT (DCD), a deep learning convolutional neural network (CNN) that uses the entire chest CT volume to automatically predict COVID-19 (COVID+) from non-COVID-19 (COVID-) pneumonia and normal controls. We discuss training strategies and differences in performance across 13 international institutions and 8 countries. The inclusion of non-China sites in training significantly improved classification performance with area under the curve (AUCs) and accuracies above 0.8 on most test sites. Furthermore, using available follow-up scans, we investigate methods to track patient disease course and predict prognosis.

Radiology artificial intelligence, a systematic evaluation of methods (RAISE): a systematic review protocol

INTRODUCTION

There has been a recent explosion of research into the field of artificial intelligence as applied to clinical radiology with the advent of highly accurate computer vision technology. These studies, however, vary significantly in design and quality. While recent guidelines have been established to advise on ethics, data management and the potential directions of future research, systematic reviews of the entire field are lacking. We aim to investigate the use of artificial intelligence as applied to radiology, to identify the clinical questions being asked, which methodological approaches are applied to these questions and trends in use over time.

METHODS AND ANALYSIS

We will follow the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines and by the Cochrane Collaboration Handbook. We will perform a literature search through MEDLINE (Pubmed), and EMBASE, a detailed data extraction of trial characteristics and a narrative synthesis of the data. There will be no language restrictions. We will take a task-centred approach rather than focusing on modality or clinical subspecialty. Sub-group analysis will be performed by segmentation tasks, identification tasks, classification tasks, pegression/prediction tasks as well as a sub-analysis for paediatric patients.

ETHICS AND DISSEMINATION

Ethical approval will not be required for this study, as data will be obtained from publicly available clinical trials. We will disseminate our results in a peer-reviewed publication. Registration number PROSPERO: CRD42020154790.

Development and Validation of a Machine Learning Model to Explore Tyrosine Kinase Inhibitor Response in Patients With Stage IV EGFR Variant-Positive Non-Small Cell Lung Cancer

IMPORTANCE

An end-to-end efficacy evaluation approach for identifying progression risk after epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (TKI) therapy in patients with stage IV EGFR variant-positive non-small cell lung cancer (NSCLC) is lacking.

OBJECTIVE

To propose a clinically applicable large-scale bidirectional generative adversarial network for predicting the efficacy of EGFR-TKI therapy in patients with NSCLC.

DESIGN, SETTING, AND PARTICIPANTS

This diagnostic/prognostic study enrolled 465 patients from January 1, 2010, to August 1, 2017, with follow-up from February 1, 2010, to June 1, 2020. A deep learning (DL) semantic signature to predict progression-free survival (PFS) was constructed in the training cohort, validated in 2 external validation and 2 control cohorts, and compared with the radiomics signature.

EXPOSURES

An end-to-end bidirectional generative adversarial network framework was designed to predict the progression risk in patients with NSCLC.

MAIN OUTCOMES AND MEASURES

The primary end point was PFS, considering the time from the initiation of therapy to the date of recurrence, confirmed disease progression, or death.

RESULTS

A total of 342 patients with stage IV EGFR variant-positive NSCLC receiving EGFR-TKI therapy met the inclusion criteria. Of these, 145 patients from 2 of the hospitals (n = 117 and 28) formed a training cohort (mean [SD] age, 61 [11] years; 87 [60.0%] female), and the patients from 2 other hospitals comprised 2 external validation cohorts (validation cohort 1: n = 101; mean [SD] age, 57 [12] years; 60 [59.4%] female; and validation cohort 2: n = 96, mean [SD] age, 58 [9] years; 55 [57.3%] female). Fifty-six patients with advanced-stage EGFR variant-positive NSCLC (mean [SD] age, 52 [11] years; 26 [46.4%] female) and 67 patients with advanced-stage EGFR wild-type NSCLC (mean [SD] age, 54 [10] years; 10 [15.0%] female) who received first-line chemotherapy were included. A total of 90 (26%) receiving EGFR-TKI therapy with a high risk of rapid disease progression were identified (median [range] PFS, 7.3 [1.4-32.0] months in the training cohort, 5.0 [0.6-34.6] months in validation cohort 1, and 6.4 [1.8-20.1] months, in validation cohort 2) using the DL semantic signature.The PFS decreased by 36% (hazard ratio, 2.13; 95% CI, 1.30-3.49; P < .001) compared with that in other patients (median [range] PFS, 11.5 [1.5-64.2] months in the training cohort, 10.9 [1.1-50.5] in validation cohort 1, and 8.9 [0.8-40.6] months in validation cohort 2. No significant differences were observed when comparing the PFS of high-risk patients receiving EGFR-TKI therapy with the chemotherapy cohorts (median PFS, 6.9 vs 4.4 months; P = .08). In terms of predicting the tumor progression risk after EGFR-TKI therapy, clinical decisions based on the DL semantic signature led to better survival outcomes than those based on radiomics signature across all risk probabilities by the decision curve analysis.

CONCLUSIONS AND RELEVANCE

This diagnostic/prognostic study provides a clinically applicable approach for identifying patients with stage IV EGFR variant-positive NSCLC who are not likely to benefit from EGFR-TKI therapy. The end-to-end DL-derived semantic features eliminated all manual interventions required while using previous radiomics methods and have a better prognostic performance.

Artificial intelligence for automatic cerebral ventricle segmentation and volume calculation: a clinical tool for the evaluation of pediatric hydrocephalus

OBJECTIVE

Imaging evaluation of the cerebral ventricles is important for clinical decision-making in pediatric hydrocephalus. Although quantitative measurements of ventricular size, over time, can facilitate objective comparison, automated tools for calculating ventricular volume are not structured for clinical use. The authors aimed to develop a fully automated deep learning (DL) model for pediatric cerebral ventricle segmentation and volume calculation for widespread clinical implementation across multiple hospitals.

METHODS

The study cohort consisted of 200 children with obstructive hydrocephalus from four pediatric hospitals, along with 199 controls. Manual ventricle segmentation and volume calculation values served as "ground truth" data. An encoder-decoder convolutional neural network architecture, in which T2-weighted MR images were used as input, automatically delineated the ventricles and output volumetric measurements. On a held-out test set, segmentation accuracy was assessed using the Dice similarity coefficient (0 to 1) and volume calculation was assessed using linear regression. Model generalizability was evaluated on an external MRI data set from a fifth hospital. The DL model performance was compared against FreeSurfer research segmentation software.

RESULTS

Model segmentation performed with an overall Dice score of 0.901 (0.946 in hydrocephalus, 0.856 in controls). The model generalized to external MR images from a fifth pediatric hospital with a Dice score of 0.926. The model was more accurate than FreeSurfer, with faster operating times (1.48 seconds per scan).

CONCLUSIONS

The authors present a DL model for automatic ventricle segmentation and volume calculation that is more accurate and rapid than currently available methods. With near-immediate volumetric output and reliable performance across institutional scanner types, this model can be adapted to the real-time clinical evaluation of hydrocephalus and improve clinician workflow.

Deep Learning for Automated Delineation of Pediatric Cerebral Arteries on Pre-operative Brain Magnetic Resonance Imaging

Introduction: Surgical resection of brain tumors is often limited by adjacent critical structures such as blood vessels. Current intraoperative navigations systems are limited; most are based on two-dimensional (2D) guidance systems that require manual segmentation of any regions of interest (ROI; eloquent structures to avoid or tumor to resect). They additionally require time- and labor-intensive processing for any reconstruction steps. We aimed to develop a deep learning model for real-time fully automated segmentation of the intracranial vessels on preoperative non-angiogram imaging sequences. Methods: We identified 48 pediatric patients (10-months to 22-years old) with high resolution (0.5-1 mm axial thickness) isovolumetric, pre-operative T2 magnetic resonance images (MRIs). Twenty-eight patients had anatomically normal brains, and 20 patients had tumors or other lesions near the skull base. Manually segmented intracranial vessels (internal carotid, middle cerebral, anterior cerebral, posterior cerebral, and basilar arteries) served as ground truth labels. Patients were divided into 80/5/15% training/validation/testing sets. A modified 2-D Unet convolutional neural network (CNN) architecture implemented with 5 layers was trained to maximize the Dice coefficient, a measure of the correct overlap between the predicted vessels and ground truth labels. Results: The model was able to delineate the intracranial vessels in a held-out test set of normal and tumor MRIs with an overall Dice coefficient of 0.75. While manual segmentation took 1-2 h per patient, model prediction took, on average, 8.3 s per patient. Conclusions: We present a deep learning model that can rapidly and automatically identify the intracranial vessels on pre-operative MRIs in patients with normal vascular anatomy and in patients with intracranial lesions. The methodology developed can be translated to other critical brain structures. This study will serve as a foundation for automated high-resolution ROI segmentation for three-dimensional (3D) modeling and integration into an augmented reality navigation platform.

Neonatal genetic epilepsies display convergent white matter microstructural abnormalities

White matter undergoes rapid development in the neonatal period. Its structure during and after development is influenced by neuronal activity. Pathological neuronal activity, as in seizures, might alter white matter, which in turn may contribute to network dysfunction. Neonatal epilepsy presents an opportunity to investigate seizures and early white matter development. Our objective was to determine whether neonatal seizures in the absence of brain injury or congenital anomalies are associated with altered white matter microstructure. In this retrospective case-control study of term neonates, cases had confirmed or suspected genetic epilepsy and normal brain magnetic resonance imaging (MRI) and no other conditions independently impacting white matter. Controls were healthy neonates with normal MRI results. White matter microstructure was assessed via quantitative mean diffusivity (MD). In 22 cases, MD was significantly lower in the genu of the corpus callosum, compared to 22 controls, controlling for gestational age and postmenstrual age at MRI. This finding suggests convergent abnormal corpus callosum microstructure in neonatal epilepsies with diverse suspected genetic causes. Further study is needed to determine the specific nature, causes, and functional impact of seizure-associated abnormal white matter in neonates, a potential pathogenic mechanism.

Diffusion tensor magnetic resonance imaging of the optic nerves in pediatric hydrocephalus

OBJECTIVE

While conventional imaging can readily identify ventricular enlargement in hydrocephalus, structural changes that underlie microscopic tissue injury might be more difficult to capture. MRI-based diffusion tensor imaging (DTI) uses properties of water motion to uncover changes in the tissue microenvironment. The authors hypothesized that DTI can identify alterations in optic nerve microstructure in children with hydrocephalus.

METHODS

The authors retrospectively reviewed 21 children (< 18 years old) who underwent DTI before and after neurosurgical intervention for acute obstructive hydrocephalus from posterior fossa tumors. Their optic nerve quantitative DTI metrics of mean diffusivity (MD) and fractional anisotropy (FA) were compared to those of 21 age-matched healthy controls.

RESULTS

Patients with hydrocephalus had increased MD and decreased FA in bilateral optic nerves, compared to controls (p < 0.001). Normalization of bilateral optic nerve MD and FA on short-term follow-up (median 1 day) after neurosurgical intervention was observed, as was near-complete recovery of MD on long-term follow-up (median 1.8 years).

CONCLUSIONS

DTI was used to demonstrate reversible alterations of optic nerve microstructure in children presenting acutely with obstructive hydrocephalus. Alterations in optic nerve MD and FA returned to near-normal levels on short- and long-term follow-up, suggesting that surgical intervention can restore optic nerve tissue microstructure. This technique is a safe, noninvasive imaging tool that quantifies alterations of neural tissue, with a potential role for evaluation of pediatric hydrocephalus.

End-to-end automatic differentiation of the coronavirus disease 2019 (COVID-19) from viral pneumonia based on chest CT

PURPOSE

In the absence of a virus nucleic acid real-time reverse transcriptase-polymerase chain reaction (RT-PCR) test and experienced radiologists, clinical diagnosis is challenging for viral pneumonia with clinical symptoms and CT signs similar to that of coronavirus disease 2019 (COVID-19). We developed an end-to-end automatic differentiation method based on CT images to identify COVID-19 pneumonia patients in real time.

METHODS

From January 18 to February 23, 2020, we conducted a retrospective study and enrolled 201 patients from two hospitals in China who underwent chest CT and RT-PCR tests, of which 98 patients tested positive for COVID-19 (118 males and 83 females, with an average age of 42 years). Patient CT images from one hospital were divided among training, validation and test datasets with an 80%:10%:10% ratio. An end-to-end representation learning method using a large-scale bi-directional generative adversarial network (BigBiGAN) architecture was designed to extract semantic features from the CT images. The semantic feature matrix was input for linear classifier construction. Patients from the other hospital were used for external validation. Differentiation accuracy was evaluated using a receiver operating characteristic curve.

RESULTS

Based on the 120-dimensional semantic features extracted by BigBiGAN from each image, the linear classifier results indicated that the area under the curve (AUC) in the training, validation and test datasets were 0.979, 0.968 and 0.972, respectively, with an average sensitivity of 92% and specificity of 91%. The AUC for external validation was 0.850, with a sensitivity of 80% and specificity of 75%. Publicly available architecture and computing resources were used throughout the study to ensure reproducibility.

CONCLUSION

This study provides an efficient recognition method for coronavirus disease 2019 pneumonia, using an end-to-end design to implement targeted and effective isolation for the containment of this communicable disease.

Artificial intelligence in stroke imaging: Current and future perspectives

Artificial intelligence (AI) is a fast-growing research area in computer science that aims to mimic cognitive processes through a number of techniques. Supervised machine learning, a subfield of AI, includes methods that can identify patterns in high-dimensional data using labeled 'ground truth' data and apply these learnt patterns to analyze, interpret, or make predictions on new datasets. Supervised machine learning has become a significant area of interest within the medical community. Radiology and neuroradiology in particular are especially well suited for application of machine learning due to the vast amount of data that is generated. One devastating disease for which neuroimaging plays a significant role in the clinical management is stroke. Within this context, AI techniques can play pivotal roles for image-based diagnosis and management of stroke. This overview focuses on the recent advances of artificial intelligence methods - particularly supervised machine learning and deep learning - with respect to workflow, image acquisition and reconstruction, and image interpretation in patients with acute stroke, while also discussing potential pitfalls and future applications.

Simultaneous time of flight-MRA and T2* imaging for cerebrovascular MRI

PURPOSE

3D multi-echo gradient-recalled echo (ME-GRE) can simultaneously generate time-of-flight magnetic resonance angiography (pTOF) in addition to T2*-based susceptibility-weighted images (SWI). We assessed the clinical performance of pTOF generated from a 3D ME-GRE acquisition compared with conventional TOF-MRA (cTOF).

METHODS

Eighty consecutive children were retrospectively identified who obtained 3D ME-GRE alongside cTOF. Two blinded readers independently assessed pTOF derived from 3D ME-GRE and compared them with cTOF. A 5-point Likert scale was used to rank lesion conspicuity and to assess for diagnostic confidence.

RESULTS

Across 80 pediatric neurovascular pathologies, a similar number of lesions were reported on pTOF and cTOF (43-40%, respectively, p > 0.05). Rating of lesion conspicuity was higher with cTOF (4.5 ± 1.0) as compared with pTOF (4.0 ± 0.7), but this was not significantly different (p = 0.06). Diagnostic confidence was rated higher with cTOF (4.8 ± 0.5) than that of pTOF (3.7 ± 0.6; p < 0.001). Overall, the inter-rater agreement between two readers for lesion count on pTOF was classified as almost perfect (κ = 0.98, 96% CI 0.8-1.0).

CONCLUSIONS

In this study, TOF-MRA simultaneously generated in addition to SWI from 3D MR-GRE can serve as a diagnostic adjunct, particularly for proximal vessel disease and when conventional TOF-MRA images are absent.

Brain Iron Assessment after Ferumoxytol-enhanced MRI in Children and Young Adults with Arteriovenous Malformations: A Case-Control Study

Background Iron oxide nanoparticles are an alternative contrast agent for MRI. Gadolinium deposition has raised safety concerns, but it is unknown whether ferumoxytol administration also deposits in the brain. Purpose To investigate whether there are signal intensity changes in the brain at multiecho gradient imaging following ferumoxytol exposure in children and young adults. Materials and Methods This retrospective case-control study included children and young adults, matched for age and sex, with brain arteriovenous malformations who received at least one dose of ferumoxytol from January 2014 to January 2018. In participants who underwent at least two brain MRI examinations (subgroup), the first and last available examinations were analyzed. Regions of interests were placed around deep gray structures on quantitative susceptibility mapping and R2* images. Mean susceptibility and R2* values of regions of interests were recorded. Measurements were assessed by linear regression analyses: a between-group comparison of ferumoxytol-exposed and unexposed participants and a within-group (subgroup) comparison before and after exposure. Results Seventeen participants (mean age ± standard deviation, 13 years ± 5; nine male) were in the ferumoxytol-exposed (case) group, 21 (mean age, 14 years ± 5; 11 male) were in the control group, and nine (mean age, 12 years ± 6; four male) were in the subgroup. The mean number of ferumoxytol administrations was 2 ± 1 (range, one to four). Mean susceptibility (in parts per million [ppm]) and R2* (in inverse seconds [sec^-1]) values of the dentate (case participants: 0.06 ppm ± 0.04 and 23.87 sec^-1 ± 4.13; control participants: 0.02 ppm ± 0.03 and 21.7 sec^-1 ± 5.26), substantia nigrae (case participants: 0.08 ppm ± 0.06 and 27.46 sec^-1 ± 5.58; control participants: 0.04 ppm ± 0.05 and 24.96 sec^-1 ± 5.3), globus pallidi (case participants: 0.14 ppm ± 0.05 and 30.75 sec^-1 ± 5.14; control participants: 0.08 ppm ± 0.07 and 28.82 sec^-1 ± 6.62), putamina (case participants: 0.03 ppm ± 0.02 and 20.63 sec^-1 ± 2.44; control participants: 0.02 ppm ± 0.02 and 19.65 sec^-1 ± 3.6), caudate (case participants: -0.1 ppm ± 0.04 and 18.21 sec^-1 ± 3.1; control participants: -0.06 ppm ± 0.05 and 18.83 sec^-1 ± 3.32), and thalami (case participants: 0 ppm ± 0.03 and 16.49 sec^-1 ± 3.6; control participants: 0.02 ppm ± 0.02 and 18.38 sec^-1 ± 2.09) did not differ between groups (susceptibility, P = .21; R2*, P = .24). For the subgroup, the mean interval between the first and last ferumoxytol administration was 14 months ± 8 (range, 1-25 months). Mean susceptibility and R2* values of the dentate (first MRI: 0.06 ppm ± 0.05 and 25.78 sec^-1 ± 5.9; last MRI: 0.06 ppm ± 0.02 and 25.55 sec^-1 ± 4.71), substantia nigrae (first MRI: 0.06 ppm ± 0.06 and 28.26 sec^-1 ± 9.56; last MRI: 0.07 ppm ± 0.06 and 25.65 sec^-1 ± 6.37), globus pallidi (first MRI: 0.13 ppm ± 0.07 and 27.53 sec^-1 ± 8.88; last MRI: 0.14 ppm ± 0.06 and 29.78 sec^-1 ± 6.54), putamina (first MRI: 0.03 ppm ± 0.03 and 19.78 sec^-1 ± 3.51; last MRI: 0.03 ppm ± 0.02 and 19.73 sec^-1 ± 3.01), caudate (first MRI: -0.09 ppm ± 0.05 and 21.38 sec^-1 ± 4.72; last MRI: -0.1 ppm ± 0.05 and 18.75 sec^-1 ± 2.68), and thalami (first MRI: 0.01 ppm ± 0.02 and 17.65 sec^-1 ± 5.16; last MRI: 0 ppm ± 0.02 and 15.32 sec^-1 ± 2.49) did not differ between the first and last MRI examinations (susceptibility, P = .95; R2*, P = .54). Conclusion No overall significant differences were found in susceptibility and R2* values of deep gray structures to suggest retained iron in the brain between ferumoxytol-exposed and unexposed children and young adults with arteriovenous malformations and in those exposed to ferumoxytol over time. © RSNA, 2020.

Deep Learning for Pediatric Posterior Fossa Tumor Detection and Classification: A Multi-Institutional Study

BACKGROUND AND PURPOSE

Posterior fossa tumors are the most common pediatric brain tumors. MR imaging is key to tumor detection, diagnosis, and therapy guidance. We sought to develop an MR imaging-based deep learning model for posterior fossa tumor detection and tumor pathology classification.

MATERIALS AND METHODS

The study cohort comprised 617 children (median age, 92 months; 56% males) from 5 pediatric institutions with posterior fossa tumors: diffuse midline glioma of the pons (n = 122), medulloblastoma (n = 272), pilocytic astrocytoma (n = 135), and ependymoma (n = 88). There were 199 controls. Tumor histology served as ground truth except for diffuse midline glioma of the pons, which was primarily diagnosed by MR imaging. A modified ResNeXt-50-32x4d architecture served as the backbone for a multitask classifier model, using T2-weighted MRIs as input to detect the presence of tumor and predict tumor class. Deep learning model performance was compared against that of 4 radiologists.

RESULTS

Model tumor detection accuracy exceeded an AUROC of 0.99 and was similar to that of 4 radiologists. Model tumor classification accuracy was 92% with an F₁ score of 0.80. The model was most accurate at predicting diffuse midline glioma of the pons, followed by pilocytic astrocytoma and medulloblastoma. Ependymoma prediction was the least accurate. Tumor type classification accuracy and F₁ score were higher than those of 2 of the 4 radiologists.

CONCLUSIONS

We present a multi-institutional deep learning model for pediatric posterior fossa tumor detection and classification with the potential to augment and improve the accuracy of radiologic diagnosis.