Quantitative assessment of cerebrospinal fluid flow and volume in enlargement of the subarachnoid spaces of infancy using MRI.

BACKGROUND: The etiology of enlarged subarachnoid spaces of infancy is unknown; however, there is radiologic similarity with normal pressure hydrocephalus. Adults with normal pressure hydrocephalus have been shown to have altered cerebrospinal (CSF) flow through the cerebral aqueduct. OBJECTIVE: To explore potential similarity between enlarged subarachnoid spaces of infancy and normal pressure hydrocephalus, we compared MRI-measured CSF flow through the cerebral aqueduct in infants with enlarged subarachnoid spaces of infancy to infants with normal brain MRIs. MATERIALS AND METHODS: This was an IRB approved retrospective study. Clinical brain MRI examinations including axial T2 imaging and phase contrast through the aqueduct were reviewed for infants with enlarged subarachnoid spaces of infancy and for infants with a qualitatively normal brain MRI. The brain and CSF volumes were segmented using a semi-automatic technique (Analyze 12.0) and CSF flow parameters were measured (cvi42, 5.14). All data was assessed for significant differences while controlling for age and sex using analysis of covariance (ANCOVA). RESULTS: Twenty-two patients with enlarged subarachnoid spaces (mean age 9.0 months, 19 males) and 15 patients with normal brain MRI (mean age 18.9 months, 8 females) were included. Volumes of the subarachnoid space (P < 0.001), lateral (P < 0.001), and third ventricles (P < 0.001) were significantly larger in infants with enlarged subarachnoid spaces of infancy. Aqueductal stroke volume significantly increased with age (P = 0.005), regardless of group. CONCLUSION: CSF volumes were significantly larger in infants with enlarged subarachnoid spaces of infancy versus infants with a normal MRI; however, there was no significant difference in CSF flow parameters between the two groups.

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.

Advanced imaging techniques for neuro-oncologic tumor diagnosis, with an emphasis on PET-MRI imaging of malignant brain tumors.

PURPOSE OF REVIEW: This review will explore the latest in advanced imaging techniques, with a focus on the complementary nature of multiparametric, multimodality imaging using magnetic resonance imaging (MRI) and positron emission tomography (PET). RECENT FINDINGS: Advanced MRI techniques including perfusion-weighted imaging (PWI), MR spectroscopy (MRS), diffusion-weighted imaging (DWI), and MR chemical exchange saturation transfer (CEST) offer significant advantages over conventional MR imaging when evaluating tumor extent, predicting grade, and assessing treatment response. PET performed in addition to advanced MRI provides complementary information regarding tumor metabolic properties, particularly when performed simultaneously. 18F-fluoroethyltyrosine (FET) PET improves the specificity of tumor diagnosis and evaluation of post-treatment changes. Incorporation of radiogenomics and machine learning methods further improve advanced imaging. The complementary nature of combining advanced imaging techniques across modalities for brain tumor imaging and incorporating technologies such as radiogenomics has the potential to reshape the landscape in neuro-oncology.

Differentiation of pilocytic and pilomyxoid astrocytomas using dynamic susceptibility contrast perfusion and diffusion weighted imaging.

PURPOSE: Pilocytic (PA) and pilomyxoid astrocytomas (PMA) are related low-grade tumors which occur predominantly in children. PMAs have a predilection for a supratentorial location in younger children with worse outcomes. However, the two have similar imaging characteristics. Quantitative MR sequences such as dynamic susceptibility contrast (DSC) perfusion and diffusion (DWI) were assessed for significant differences between the two tumor types and locations. METHODS: A retrospective search for MRI with DSC and DWI on pathology-proven cases of PMA and PA in children was performed. Tumors were manually segmented on anatomic images registered to rCBV, K2, and ADC maps. Tumors were categorized as PA or PMA, with subclassification of supratentorial and infratentorial locations. Mean values were obtained for tumor groups and locations compared with Student's t test for significant differences with post hoc correction for multiple comparisons. ROC analysis for significant t test values was performed. Histogram evaluation was also performed. RESULTS: A total of 49 patients met inclusion criteria. This included 30 patients with infratentorial PA, 8 with supratentorial PA, 6 with supratentorial PMA, and 5 with infratentorial PMA. Mean analysis showed significantly increased rCBV for infratentorial PMA (2.39 ± 1.1) vs PA (1.39 ± 0.16, p = 0.0006). ROC analysis for infratentorial PA vs PMA yielded AUC = 0.87 (p < 0.001). Histogram analysis also demonstrated a higher ADC peak location for PMA (1.8 ± 0.2) vs PA (1.56 ± 0.28). CONCLUSION: PMA has a significantly higher rCBV than PA in the infratentorial space. DSC perfusion and diffusion MR imaging may be helpful to distinguish between the two tumor types in this location.

Black bone MRI with 3D reconstruction for the detection of skull fractures in children with suspected abusive head trauma.

PURPOSE: The purpose of this study was to determine the accuracy of "black bone" (BB) MRI for the detection of skull fractures in children with potential abusive head trauma. METHODS: A total of 34 pediatric patients were evaluated for potential abusive head trauma. All patients had both a non-contrast head CT (HCT) with multiplanar reformatted images and 3D volumetric reformatted images where available (gold standard) for fracture diagnosis and BB of the head with multiplanar reformatted images and 3D volumetric images. BB was performed using an ultrashort TE pointwise encoding time reduction with radial acquisition (PETRA) sequence at 1.5 T or 3 T. BB datasets were post-processed and 3D images created using Fovia's High Definition Volume Rendering® software. Two board-certified pediatric neuroradiologists independently reviewed the HCT and BB imaging, blinded to the findings from the other modality. RESULTS: Median patient age was 4 months (range 1.2-30 months). A total of 20 skull fractures in six patients (18% incidence of skull fractures) were detected on HCT. BB demonstrated 83% sensitivity (95%[CI] 36-99%), 100% specificity (95%[CI] 88-100%), 100% PPV (95%[CI] 46-100%), 97% NPV (95%[CI] 82-99%), and 97% accuracy (95%[CI] 85-99%) for diagnosis of a skull fracture. BB detected 95% (19/20) of the skull fractures detected by CT. CONCLUSION: A black bone MRI sequence may provide high sensitivity and specificity for detection of skull fractures in pediatric patients with abusive head trauma.

Comparison of multi-shot and single shot echo-planar diffusion tensor techniques for the optic pathway in patients with neurofibromatosis type 1.

PURPOSE: Diffusion tensor imaging (DTI) may be helpful in assessing optic pathway integrity as a marker for treatment in neurofibromatosis type 1 (NF1) patients with optic gliomas (OG). However, susceptibility artifacts are common in typical single-shot echo planar imaging (ssDTI). A readout-segmented multi-shot EPI technique (rsDTI) was utilized to minimize susceptibility distortions of the skull base and improve quantitative metrics. METHODS: Healthy controls, children with NF1 without OG, and NF1 with OG ± visual symptoms were included. All subjects were scanned with both rsDTI and ssDTI sequences sequentially. Diffusion metrics and deterministic fiber tracking were calculated. Tract count, volume, and length were also compared by a two-factor mixed ANOVA. RESULTS: Five healthy controls, 7 NF1 children without OG, and 12 NF1 children with OG were imaged. Six OG patients had visual symptoms. Four subjects had no detectable optic pathway fibers on ssDTI due to susceptibility, for which rsDTI was able to delineate. Tract count (p < 0.001), tract volume (p < 0.001), and FA (P < 0.001) were significantly higher for rsDTI versus ssDTI for all subjects. MD (p < 0.001) and RD (p < 0.001) were significantly lower for rsDTI vs ssDTI. Finally, MD, AD, and RD had a significantly lower difference in NF1 children with visual symptoms compared to NF1 children without visual symptoms only on ssDTI scans. CONCLUSION: DTI with readout-segmented multi-shot EPI technique can better visualize the optic pathway and allow more confident measurements of anisotropy in NF1 patients. This is shown by a significant increase in FA, tract count, and volume with rsDTI versus ssDTI.

Radiologic head CT interpretation errors in pediatric abusive and non-abusive head trauma patients.

BACKGROUND: Pediatric head trauma, including abusive head trauma, is a significant cause of morbidity and mortality. OBJECTIVE: The purpose of this research was to identify and evaluate radiologic interpretation errors of head CTs performed on abusive and non-abusive pediatric head trauma patients from a community setting referred for a secondary interpretation at a tertiary pediatric hospital. MATERIALS AND METHODS: A retrospective search identified 184 patients <5 years of age with head CT for known or potential head trauma who had a primary interpretation performed at a referring community hospital by a board-certified radiologist. Two board-certified fellowship-trained neuroradiologists at an academic pediatric hospital independently interpreted the head CTs, compared their interpretations to determine inter-reader discrepancy rates, and resolved discrepancies to establish a consensus second interpretation. The primary interpretation was compared to the consensus second interpretation using the RADPEER™ scoring system to determine the primary interpretation-second interpretation overall and major discrepancy rates. MRI and/or surgical findings were used to validate the primary interpretation or second interpretation when possible. The diagnosis of abusive head trauma was made using clinical and imaging data by a child abuse specialist to separate patients into abusive head trauma and non-abusive head trauma groups. Discrepancy rates were compared for both groups. Lastly, primary interpretations and second interpretations were evaluated for discussion of imaging findings concerning for abusive head trauma. RESULTS: There were statistically significant differences between primary interpretation-second interpretation versus inter-reader overall and major discrepancy rates (28% vs. 6%, P=0.0001; 16% vs. 1%, P=0.0001). There were significant differences in the primary interpretation-second interpretation overall and major discrepancy rates for abusive head trauma patients compared to non-abusive head trauma patients (41% vs 23%, P=0.02; 26% vs. 12%, P=0.03). The most common findings resulting in major radiologic interpretation errors were fractures and subdural hemorrhage. Differences in the age of the patient and the percentage of patients with hemorrhage were statistically significant between the abusive head trauma versus non-abusive head trauma groups, while no statistical difference was identified for skull fractures, ischemia, head CT radiation dose, or presence of multiplanar or 3-D reformatted images. The second interpretation more frequently indicated potential for abusive head trauma compared to the primary interpretation (P=0.0001). MRI and/or surgical findings were in agreement with the second interpretation in 29/29 (100%) of patients with discrepancies. CONCLUSION: A high incidence of radiologic interpretation errors may occur in pediatric trauma patients at risk for abusive head trauma who are referred from a community hospital. This suggests value for second interpretations of head CTs at a tertiary pediatric hospital for this patient population.

Relative cerebral blood volume from dynamic susceptibility contrast perfusion in the grading of pediatric primary brain tumors.

INTRODUCTION: The aim of this study is to evaluate the utility of relative cerebral blood volume (rCBV) data from dynamic susceptibility contrast (DSC) perfusion in grading pediatric primary brain tumors. METHODS: A retrospective blinded review of 63 pediatric brain tumors with DSC perfusion was performed independently by two neuroradiologists. A diagnosis of low- versus high-grade tumor was obtained from conventional imaging alone. Maximum rCBV (rCBVmax) was measured from manual ROI placement for each reviewer and averaged. Whole-tumor CBV data was obtained from a semi-automated approach. Results from all three analyses were compared to WHO grade. RESULTS: Based on conventional MRI, the two reviewers had a concordance rate of 81% (k = 0.62). Compared to WHO grade, the concordant cases accurately diagnosed high versus low grade in 82%. A positive correlation was demonstrated between manual rCBVmax and tumor grade (r = 0.30, P = 0.015). ROC analysis of rCBVmax (area under curve 0.65, 0.52-0.77, P = 0.03) gave a low-high threshold of 1.38 with sensitivity of 92% (74-99%), specificity of 40% (24-57%), NPV of 88% (62-98%), and PPV of 50% (35-65%) Using this threshold on 12 discordant tumors between evaluators from conventional imaging yielded correct diagnoses in nine patients. Semi-automated analysis demonstrated statistically significant differences between low- and high-grade tumors for multiple metrics including average rCBV (P = 0.027). CONCLUSIONS: Despite significant positive correlation with tumor grade, rCBV from pediatric brain tumors demonstrates limited specificity, but high NPV in excluding high-grade neoplasms. In selective patients whose conventional imaging is nonspecific, an rCBV threshold may have further diagnostic value.

Ectopic external auditory canal and ossicular formation in the oculo-auriculo-vertebral spectrum.

Ear abnormalities in oculo-auricular-vertebral spectrum commonly present with varying degrees of external and middle ear atresias, usually in the expected locations of the temporal bone and associated soft tissues, without ectopia of the external auditory canal. We present the unique imaging of a 4-year-old girl with right hemifacial microsomia and ectopic location of an atretic external auditory canal, terminating in a hypoplastic temporomandibular joint containing bony structures with the appearance of auditory ossicles. This finding suggests an early embryological dysfunction involving Meckel's cartilage of the first branchial arch.

Development of Myelin Growth Charts of the White Matter Using T1 Relaxometry.

BACKGROUND AND PURPOSE: Myelin maturation occurs in late fetal life to early adulthood, with the most rapid changes observed in the first few years of infancy. To quantify the degree of myelination, a specific MR imaging sequence is required to measure the changes in tissue proton relaxivity (R1). R1 positively correlates with the degree of myelination maturation at a given age. Similar to head circumference charts, these data can be used to develop normal growth charts for specific white matter tracts to detect pathologies involving abnormal myelination. MATERIALS AND METHODS: This is a cross-sectional study using normal clinical pediatric brain MR images with the MP2RAGE sequence to generate T1 maps. The T1 maps were segmented to 75 brain regions from a brain atlas (white matter and gyri). Statistical modeling for all subjects across regions and the age range was computed, and estimates of population-level percentile ranking were computed to describe the effective myelination rate as a function of age. Test-retest analysis was performed to assess reproducibility. Logistic trendline and regression were performed for selected white matter regions and plotted for growth charts. RESULTS: After exclusion for abnormal MR imaging or diseases affecting myelination from the electronic medical record, 103 subject MR images were included, ranging from birth to 17 years of age. Test-retest analysis resulted in a high correlation for white matter (r = 0.88) and gyri (r = 0.95). All white matter regions from the atlas had significant P values for logistic regression with R 2 values ranging from 0.41 to 0.99. CONCLUSIONS: These data can serve as a myelination growth chart to permit patient comparisons with normal levels with respect to age and brain regions, thus improving detection of developmental disorders affecting myelin.

Accuracy of CT cerebral perfusion in predicting infarct in the emergency department: lesion characterization on CT perfusion based on commercially available software.

This study aims to assess the diagnostic accuracy of a single vendor commercially available CT perfusion (CTP) software in predicting stroke. A retrospective analysis on patients presenting with stroke-like symptoms within 6 h with CTP and diffusion-weighted imaging (DWI) was performed. Lesion maps, which overlays areas of computer-detected abnormally elevated mean transit time (MTT) and decreased cerebral blood volume (CBV), were assessed from a commercially available software package and compared to qualitative interpretation of color maps. Using DWI as the gold standard, parameters of diagnostic accuracy were calculated. Point biserial correlation was performed to assess for relationship of lesion size to a true positive result. Sixty-five patients (41 females and 24 males, age range 22-92 years, mean 57) were included in the study. Twenty-two (34 %) had infarcts on DWI. Sensitivity (83 vs. 70 %), specificity (21 vs. 69 %), negative predictive value (77 vs. 84 %), and positive predictive value (29 vs. 50 %) for lesion maps were contrasted to qualitative interpretation of perfusion color maps, respectively. By using the lesion maps to exclude lesions detected qualitatively on color maps, specificity improved (80 %). Point biserial correlation for computer-generated lesions (R pb = 0.46, p < 0.0001) and lesions detected qualitatively (R pb = 0.32, p = 0.0016) demonstrated positive correlation between size and infarction. Seventy-three percent (p = 0.018) of lesions which demonstrated an increasing size from CBV, cerebral blood flow, to MTT/time to peak were true positive. Used in isolation, computer-generated lesion maps in CTP provide limited diagnostic utility in predicting infarct, due to their inherently low specificity. However, when used in conjunction with qualitative perfusion color map assessment, the lesion maps can help improve specificity.

Imatinib mesylate for plexiform neurofibromas in patients with neurofibromatosis type 1: a phase 2 trial.

BACKGROUND: Plexiform neurofibromas are slow-growing chemoradiotherapy-resistant tumours arising in patients with neurofibromatosis type 1 (NF1). Currently, there are no viable therapeutic options for patients with plexiform neurofibromas that cannot be surgically removed because of their proximity to vital body structures. We undertook an open-label phase 2 trial to test whether treatment with imatinib mesylate can decrease the volume burden of clinically significant plexiform neurofibromas in patients with NF1. METHODS: Eligible patients had to be aged 3-65 years, and to have NF1 and a clinically significant plexiform neurofibroma. Patients were treated with daily oral imatinib mesylate at 220 mg/m(2) twice a day for children and 400 mg twice a day for adults for 6 months. The primary endpoint was a 20% or more reduction in plexiform size by sequential volumetric MRI imaging. Clinical data were analysed on an intention-to-treat basis; a secondary analysis was also done for those patients able to take imatinib mesylate for 6 months. This trial is registered with ClinicalTrials.gov, number NCT01673009. FINDINGS: Six of 36 patients (17%, 95% CI 6-33), enrolled on an intention-to-treat basis, had an objective response to imatinib mesylate, with a 20% or more decrease in tumour volume. Of the 23 patients who received imatinib mesylate for at least 6 months, six (26%, 95% CI 10-48) had a 20% or more decrease in volume of one or more plexiform tumours. The most common adverse events were skin rash (five patients) and oedema with weight gain (six). More serious adverse events included reversible grade 3 neutropenia (two), grade 4 hyperglycaemia (one), and grade 4 increases in aminotransferase concentrations (one). INTERPRETATION: Imatinib mesylate could be used to treat plexiform neurofibromas in patients with NF1. A multi-institutional clinical trial is warranted to confirm these results. FUNDING: Novartis Pharmaceuticals, the Indiana University Simon Cancer Centre, and the Indiana University Herman B Wells Center for Pediatric Research.

Temporal and optic pathway pilomyxoid astrocytoma mimicking dural-based lesion: case report and review of the literature.

Pilomyxoid astrocytomas (PMAs) are low-grade tumors that share many common traits with pilocytic astrocytomas. However, PMAs have a more worrisome clinical course, with a higher recurrence rate, lower survival rate, and higher risk of leptomeningeal spread compared to pilocytic tumors. These tumors tend to occur in younger children and are typically located in the area of the optic chiasm or hypothalamus. There are few studies examining the radiographic appearance of these lesions. In this case report, the authors present an unusual radiographic appearance of a PMA in an 11-year-old child. Preoperative images suggested a dural-based, homogenously enhancing lesion coupled with an enlarged optic nerve. Surgery revealed an intraparenchymal lesion of the right temporal lobe. There was hyperintensity on T2 MRI sequences, suggesting infiltration of the tumor along the optic tracts.

COVID-19 Associated leukoencephalopathy in a term neonate: imaging findings and clinical presentation.

A 5-day old neonate presented with several episodes of seizure-like activity associated with hypoxia. The episodes were responsive to anti-epileptic medications and the infant was given empiric antibiotics and antiviral coverage. Cerebrospinal fluid polymerase chain reaction (PCR), culture, and gram stain were negative for viral or bacterial etiology. However, a nasopharyngeal PCR of the infant was positive for SARS-COV-2. While head computed tomography (CT) was negative, magnetic resonance imaging (MRI) showed evidence of white matter injury in the subcortical and periventricular regions and corpus callosum. With supportive therapies, the infant made a full neurologic recovery and was discharged following a 5-day admission. This case highlights the growing evidence of SARS-COV-2 associated leukoencephalopathy in neonates, and physicians should consider this diagnosis in neonates with similar presentation.

Schwannoma-associated third nerve palsy: A pediatric case report.

Acquired third nerve palsies are infrequently seen in children, but are often associated with serious pathologies. This article presents a pediatric case of tumor-associated, isolated third nerve palsy, which took two years to diagnose. The patient initially presented with an isolated, dilated pupil and progressed over several months to a complete third nerve palsy. In this case, high-resolution neuroimaging eventually led to the diagnosis of a presumed schwannoma as the cause of her third nerve palsy. We review her case, the importance of high-resolution imaging, and management options.

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 < .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 = .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.

Longitudinal MRI brain findings in the R1349Q pathogenic variant of CACNA1A.

Pathogenic CACNA1A gene variants are associated with a spectrum of disorders including migraine with or without hemiplegia, ataxia, epilepsy, and developmental disability. We present a case of a pathogenic variant (c.4046G>A, p.R1349Q) in the CACNA1A gene associated with a clinical phenotype of global developmental delay, left hemiparesis, epilepsy, and stroke-like episodes. Longitudinal neuroimaging demonstrates hemispheric encephalomalacia with mismatched perfusion and angiographic imaging, in addition to progressive cerebellar atrophy.

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.

Integrating neuroimaging biomarkers into the multicentre, high-dose erythropoietin for asphyxia and encephalopathy (HEAL) trial: rationale, protocol and harmonisation.

INTRODUCTION: MRI and MR spectroscopy (MRS) provide early biomarkers of brain injury and treatment response in neonates with hypoxic-ischaemic encephalopathy). Still, there are challenges to incorporating neuroimaging biomarkers into multisite randomised controlled trials. In this paper, we provide the rationale for incorporating MRI and MRS biomarkers into the multisite, phase III high-dose erythropoietin for asphyxia and encephalopathy (HEAL) Trial, the MRI/S protocol and describe the strategies used for harmonisation across multiple MRI platforms. METHODS AND ANALYSIS: Neonates with moderate or severe encephalopathy enrolled in the multisite HEAL trial undergo MRI and MRS between 96 and 144 hours of age using standardised neuroimaging protocols. MRI and MRS data are processed centrally and used to determine a brain injury score and quantitative measures of lactate and n-acetylaspartate. Harmonisation is achieved through standardisation-thereby reducing intrasite and intersite variance, real-time quality assurance monitoring and phantom scans. ETHICS AND DISSEMINATION: IRB approval was obtained at each participating site and written consent obtained from parents prior to participation in HEAL. Additional oversight is provided by an National Institutes of Health-appointed data safety monitoring board and medical monitor. TRIAL REGISTRATION NUMBER: NCT02811263; Pre-result.

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.