An international study presenting a federated learning AI platform for pediatric brain tumors.

While multiple factors impact disease, artificial intelligence (AI) studies in medicine often use small, non-diverse patient cohorts due to data sharing and privacy issues. Federated learning (FL) has emerged as a solution, enabling training across hospitals without direct data sharing. Here, we present FL-PedBrain, an FL platform for pediatric posterior fossa brain tumors, and evaluate its performance on a diverse, realistic, multi-center cohort. Pediatric brain tumors were targeted due to the scarcity of such datasets, even in tertiary care hospitals. Our platform orchestrates federated training for joint tumor classification and segmentation across 19 international sites. FL-PedBrain exhibits less than a 1.5% decrease in classification and a 3% reduction in segmentation performance compared to centralized data training. FL boosts segmentation performance by 20 to 30% on three external, out-of-network sites. Finally, we explore the sources of data heterogeneity and examine FL robustness in real-world scenarios with data imbalances.

International Consensus Statement on the Radiological Evaluation of Dysraphic Malformations of the Spine and Spinal Cord.

Dysraphic malformations of the spine and spinal cord (DMSSC) represent a spectrum of common congenital anomalies typically (though not exclusively) affecting the lower spinal segments. These may be responsible for varying degrees of neurologic, orthopedic, and urologic morbidity. With advances in neuroimaging, it is now possible to better diagnose and evaluate these disorders both prenatally and postnatally. Neuroimaging, performed at the right time and with technique optimization, is integral in guiding clinical management. However, the terminology used to describe these lesions has become increasingly confusing, and there is a lack of consensus regarding the essential radiologic features and their clinical weighting. This variability in radiologic practice risks unstructured decision making and increases the likelihood of suboptimal, less informed clinical management. In this manuscript, the first of a series of consensus statements, we outline a standardized international consensus statement for the radiologic evaluation of children with suspected DMSSC derived from a critical review of the literature, and the collective clinical experience of a multinational group of experts. We provide recommendations for plain radiography, sonography, CT, and MR imaging in the evaluation of DMSSC with an emphasis on technique of imaging and imaging protocols.

Transient Ischemic Attack (TIA) presenting with isolated anomic aphasia due to ischemia in the posterior cingulate gyrus in a 11-year-old girl.

INTRODUCTION: Anomic aphasia, characterized by difficulty in word finding, is a subtype without impairments in fluent speech, comprehension, reading, writing, and repetition. Recognizing pure anomic aphasia in this group is crucial for a comprehensive understanding of localization and brain functions. CASE REPORT: We present the case of an 11-year-old girl with transient ischemic attack and anomic aphasia. Neuroimaging identified abnormalities in the cingulate gyrus and temporo-occipital regions. No focal neurological findings were observed without aphasia. In terms of etiology, an MTHFR mutation was detected, and aside from this, no hematological or systemic cause could be identified. CONCLUSION: This case marks the first demonstration of posterior cingulate gyrus involvement in pure anomic aphasia.

¿Podría contribuir la resonancia magnética a detectar la ventriculomegalia fetal aislada aparte de otras anormalidades?

Objetivo. Determinar el papel de la resonancia magnética (RM) cerebral en fetos que presentan ventriculomegalia aislada (VMA) en la evaluación ecográfica del cerebro fetal. Métodos. Se evaluaron retrospectivamente los hallazgos por ecografía y RM de 197 fetos diagnosticados con VMA entre noviembre de 2018 y noviembre de 2020. Se excluyeron los fetos con cariotipos anormales, anomalías adicionales o etiologías relacionadas a ventriculomegalia. Se comparó los resultados de ecografía y RM tanto en términos de medidas ventriculares medias como de grado de VMA. Resultados. Las mediciones de la RM fueron significativamente mayores en la VMA leve (10,33±0,38 mm frente a 11,11±0,51 mm, p<0,001) en comparación con la ecografía. En la VMA leve, la RM midió los ventrículos más anchos que la ecografía, con una diferencia media de 0,78 mm. No hubo diferencias significativas en las mediciones por ecografía y RM en cuanto a los valores medios de la VMA moderada y grave. Hubo buena concordancia entre la ecografía y la RM en la detección de la gravedad de la VMA derecha, izquierda y la media (Κ=0,265, Κ=0,324 y Κ=0,261, respectivamente). Los análisis de regresión lineal revelaron una relación estadísticamente significativa entre las mediciones de ecografía y RM de la VMA derecha, izquierda y la media (p<0,001, p<0,001 y p<0,001, respectivamente). La RM mostró una concordancia perfecta con la ecografía en detectar la lateralidad de la VMA (Κ=1,0, p<0,001). Conclusiones. En fetos con VMA leve detectada por ecografía se debe considerar la evaluación por RM del cerebro fetal para un diagnóstico preciso. Este enfoque puede proporcionar una estrategia eficaz en el manejo prenatal y el asesoramiento de estos embarazos.

Objective: To assess the role of brain magnetic resonance imaging (MRI) in fetuses presenting with isolated ventriculomegaly (IVM) in the ultrasound (US) evaluation of the fetal brain. Methods: US and MRI findings of 197 fetuses diagnosed with IVM between November 2018 and November 2020 were retrospectively evaluated. Fetuses with abnormal karyotypes, additional anomalies, or known etiologies for ventriculomegaly were excluded. US and MRI findings were compared both in terms of mean ventricular measurements and IVM grade. Results: MRI measurements were significantly higher in mild IMV (10.33 ± 0.38 mm vs. 11.11 ± 0.51 mm, p< 0.001) compared to US. In mild IVM, MRI measured ventricles larger than US with a mean difference of 0.78 mm. There was no significant difference in US and MRI measurements in terms of mean values in moderate and severe IVM. There was good agreement between US and MRI in detecting right, left and mean IVM severity (Κ=0.265, Κ=0.324, and Κ=0.261, respectively). Linear regression analyses revealed a statistically significant relationship between US and MRI measurements of the right, left, and mean IVM (p<0.001, p<0.001, and p<0.001, respectively). MRI showed perfect agreement with US in detecting IVM laterality (Κ=1.0, p<0.001). Conclusions: In fetuses with mild IVM detected by US, fetal brain MRI evaluation should be considered for accurate diagnosis. This approach may provide effective strategies in the antenatal management and counseling of these pregnancies.

An algorithmic clinicoradiological approach to paediatric cranial vault lesions: distinguishing normal variants from pathologies.

Lesions of the paediatric cranial vault are diverse both in their presentation and aetiology. As such, they pose a diagnostic challenge to the paediatric neurosurgeon and neuroradiologist. In this article, we delineate the spectrum of paediatric calvarial pathology into four distinct groups: (1) lytic lesion(s); (2) focal sclerotic lesion(s); (3) diffuse cranial vault sclerosis; and (4) abnormal shape of the cranial vault. It is our aim that this more pragmatic, algorithmic approach may mitigate diagnostic uncertainty and aid the more accurate diagnosis of paediatric calvarial lesions.

Contribution of fetal magnetic resonance imaging in the evaluation of neurosonographically detected cases of isolated mild and moderate cerebral ventriculomegaly.

AIM: This study aimed to present the contribution of prenatal magnetic resonance imaging (MRI) in the diagnosis of fetuses that were previously identified as isolated mild and moderate cerebral ventriculomegaly (VM) by ultrasound (US). METHODS: The data between February 2013 and August 2020 were collected for women who were diagnosed with isolated mild or moderate fetal VM by US and subsequently underwent a fetal MRI. RESULTS: Among 321 women, 21 (6.5%) had a clinically important additional finding after MRI. Twelve of 276 (4.3%) fetuses with mild VM and 9 of 45 (20%) with moderate VM had turned out to have additional central nervous system abnormalities. Additional findings were detected more in fetuses with moderate VM, mothers with an anterior-located placenta, and mothers with higher body mass indexes (BMIs) with statistical significance (p = 0.001, p = 0.013, p = 0.036, respectively). The most common additional MRI finding was grade 3 or 4 germinal matrix hemorrhage, which was detected in 11 of 21 fetuses (52.3%). CONCLUSIONS: Considering the countries' health policies, prenatal MRI would contribute mostly to the diagnosis of fetuses with moderate VM, pregnancies with anterior-located placenta, and mothers with high BMIs. According to our data, we believe that MRI will be valuable, especially in the diagnosis of grade 3 and 4 intracranial hemorrhage group.

Fetal corpus callosum abnormalities: Ultrasound and magnetic resonance imaging role.

The corpus callosum (CC) is the major interhemispheric commissure and its abnormalities include agenesis, hypoplasia, and hyperplasia. The CC anomalies are typically related to other central nervous system (CNS) or extra-CNS malformations. The antenatal diagnosis of complete CC agenesis is easy after mid-trimester by ultrasound (US) even in the axial plane. The non-visualization of cavum septum pellucidum and colpocephaly are critical signs in the axial view. More subtle findings (i.e., hypoplasia and partial agenesis) might also be recognized antenatally. In this review, the focus was given on the prenatal diagnosis of CC abnormalities in US and magnetic resonance imaging.

Attention-guided deep learning for gestational age prediction using fetal brain MRI.

Magnetic resonance imaging offers unrivaled visualization of the fetal brain, forming the basis for establishing age-specific morphologic milestones. However, gauging age-appropriate neural development remains a difficult task due to the constantly changing appearance of the fetal brain, variable image quality, and frequent motion artifacts. Here we present an end-to-end, attention-guided deep learning model that predicts gestational age with R2 score of 0.945, mean absolute error of 6.7 days, and concordance correlation coefficient of 0.970. The convolutional neural network was trained on a heterogeneous dataset of 741 developmentally normal fetal brain images ranging from 19 to 39 weeks in gestational age. We also demonstrate model performance and generalizability using independent datasets from four academic institutions across the U.S. and Turkey with R2 scores of 0.81-0.90 after minimal fine-tuning. The proposed regression algorithm provides an automated machine-enabled tool with the potential to better characterize in utero neurodevelopment and guide real-time gestational age estimation after the first trimester.

Neuroradiological Mimics of Periventricular Leukomalacia.

AIM: Periventricular leukomalacia (PVL) is a term reserved to describe white matter injury in the premature brain. In this review article, the authors highlight the common and rare pathologies mimicking the chronic stage of PVL and propose practical clinico-radiological criteria that would aid in diagnosis and management. METHODS AND RESULTS: The authors first describe the typical brain MRI (magnetic resonance imaging) features of PVL. Based on their clinical presentation, pathologic entities and their neuroimaging findings were clustered into distinct categories. Three clinical subgroups were identified: healthy children, children with stable/nonprogressive neurological disorder, and those with progressive neurological disorder. The neuroradiological discriminators are described in each subgroup with relevant differential diagnoses. The mimics were broadly classified into normal variants, acquired, and inherited disorders. CONCLUSIONS: The term "PVL" should be used appropriately as it reflects its pathomechanism. The phrase "white matter injury of prematurity" or "brain injury of prematurity" is more specific. Discrepancies in imaging and clinical presentation must be tread with caution and warrant further investigations to exclude other possibilities.

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.

Evidence of disrupted rhombic lip development in the pathogenesis of Dandy-Walker malformation.

Dandy-Walker malformation (DWM) and Cerebellar vermis hypoplasia (CVH) are commonly recognized human cerebellar malformations diagnosed following ultrasound and antenatal or postnatal MRI. Specific radiological criteria are used to distinguish them, yet little is known about their differential developmental disease mechanisms. We acquired prenatal cases diagnosed as DWM and CVH and studied cerebellar morphobiometry followed by histological and immunohistochemical analyses. This was supplemented by laser capture microdissection and RNA-sequencing of the cerebellar rhombic lip, a transient progenitor zone, to assess the altered transcriptome of DWM vs control samples. Our radiological findings confirm that the cases studied fall within the accepted biometric range of DWM. Our histopathological analysis points to reduced foliation and inferior vermian hypoplasia as common features in all examined DWM cases. We also find that the rhombic lip, a dorsal stem cell zone that drives the growth and maintenance of the posterior vermis is specifically disrupted in DWM, with reduced proliferation and self-renewal of the progenitor pool, and altered vasculature, all confirmed by transcriptomics analysis. We propose a unified model for the developmental pathogenesis of DWM. We hypothesize that rhombic lip development is disrupted through either aberrant vascularization and/or direct insult which causes reduced proliferation and failed expansion of the rhombic lip progenitor pool leading to disproportionate hypoplasia and dysplasia of the inferior vermis. Timing of insult to the developing rhombic lip (before or after 14 PCW) dictates the extent of hypoplasia and distinguishes DWM from CVH.

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.

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.

Evaluation of the relationship between cranial magnetic resonance imaging findings and clinical status in children with cerebral palsy

Background/aim: The objective of this study was to evaluate the relationship between cranial magnetic resonance imaging (MRI) findings and clinical features in cerebral palsy (CP). Materials and methods: Children aged 3 to 18 years, who were followed with the diagnosis of CP between January 2012 and September 2015, were included. The type of CP was classified using the European Cerebral Palsy Monitoring Group's classification system and then, patients were divided into two groups as spastic or nonspastic groups. The Gross Motor Function Classification System (GMFCS) was used to determine the level of mobility. According to the GMFCS, levels 1, 2, and 3 were grouped as mobile, and levels 4 and 5 were grouped as immobile. Cranial MRI findings were reevaluated by a voluntarily radiologist and grouped as periventricular leukomalacia (PVL) (grades 1, 2, and 3), cerebral atrophy, migration anomaly, cerebellar involvement, basal ganglion involvement, and normal MRI findings. Results: Sixty-two patients were enrolled. The rate of mobile patients did not differ between the spastic and nonspastic groups. The incidence of PVL was significantly higher in cases of prematurity and spastic CP (p < 0.05). The rate of mobilization was significantly lower and the rate of epilepsy was significantly higher in patients with PVL. Immobile patients were more common among cases of grade 3 PVL (p < 0.05). Conclusion: The most common cranial MRI pathology was PVL, and the presence of PVL and its grade might help clinically assess the patient's CP type and level of mobilization. While pathology was observed mostly in cranial MRI in cases of CP with similar clinical features, the fact that cranial MRI was completely normal for 14.5% of the cases suggests that there may be some pathologies that we could not identify with today's imaging technology.

A Neurological Appearance of Celiac Disease: Is There Any Associated Factor?

OBJECTIVE: Celiac disease may present with one or more neurological signs and/or symptoms. We aimed to define the incidence of accompanying neurological manifestations in children diagnosed as having celiac disease. METHODS: The prospective study included 146 children diagnosed as having celiac disease. The medical records (presentation symptoms, clinical findings, serological test, duodenal biopsy results, lack/deficiency of vitamin, tissue type, accompanying autoimmune disorders) and demographic data of all patients were also reviewed. RESULTS: Thirty-five (23.9%) of the 146 celiac patients exhibited one or more neurological findings. Headache (11.6%) and dizziness (6.1%) were the most common symptoms among neurological manifestations. There was a significant difference between the patients with and without neurological manifestations in terms of sex, biopsy result, and tissue type (P < 0.05). Moreover, there was a statistically significant difference between tissue types of the patients with and without headache (P < 0.05). We found that grade 3a by Marsh classification was the most common type among the patients with and without neurological findings in celiac disease. On neuroimaging evaluation of patients, 1 patient with chronic focal ischemic lesion, 1 patient with Chiari type 1 malformation, and 1 patient with subcortical white matter changes were identified. CONCLUSIONS: Pathophysiology of neurological involvement in celiac disease is liable for various neurological findings. This study contributes to data suggesting that female sex, mild histopathological form, and human leukocyte antigen DQ2 heterozygosity are related to neurological manifestations, and also human leukocyte antigen DQ2 heterozygosity is associated with headache in celiac disease.

Topiramate-Responsive Tremor in a Novel Pathogenic Variant of SPG15 Patient.

Hereditary spastic paraplegia (HSP) is group of a rare neurodegenerative disorder with both genetically and clinically diverse neurologic features. Indeed, disease progression is varying greatly within the different forms and current treatment modalities are exclusively symptomatic for HSP. Tremor in HSP patients is only mentioned with rare case reports, so treatment option is lack in clinical ground. We reported a case of a HSP-15 girl with a previously reported novel mutation of SPG15 complained of a life disturbing tremor and topiramate as a drug therapy for tremor in our HSP patient.

Pediatric case of persistent hiccups associated with hypertrophic olivary degeneration.

BACKGROUND: Hypertrophic olivary degeneration (HOD) is a rare degenerative disorder that is thought to occur subsequent to a disruption of the dentate-rubro-olivary pathway. CASE: We report a pediatric case of unilateral HOD presented with persistent hiccups and palatal tremor. Radiological examination of diaphragm was normal considering ultrasound and chest x-ray. On T2WI (weighted images) and Fluid Attenuated Inversion Recovery (FLAIR) images, hyperintense enlargement of the right inferior olivary nucleus was seen. No abnormal enhancement was detected on post-contrast scans and no evidence of restricted diffusion was seen. Susceptibility weighted imaging (SWI) sequences revealed a chronic hemorrhage involving the medulla oblongata and cerebellum. Cranial magnetic resonance imaging (MRI) findings were consistent with unilateral HOD. Palatal tremor and dentate-rubral tremor are frequent presentation of HOD, however to our knowledge persistent hiccups had not yet been reported in children with HOD. CONCLUSION: We highlight a pediatric case of unilateral HOD, which presented with persistent hiccups. Awareness of clinical and radiological findings of HOD is important to avoid misinterpretation as a mass lesion, an ischemic event, or a demyelinating disease and provide adequate management.

Correction to: Venous pathologies in paediatric neuroradiology: from foetal to adolescent life.

The original version of this article unfortunately contained a referencing omission. Figure 11 is reused from the original publication of Figure 10 of Gunny and Lin [1].

Venous pathologies in paediatric neuroradiology: from foetal to adolescent life.

The interpretation of cerebral venous pathologies in paediatric practice is challenging as there are several normal anatomical variants, and the pathologies are diverse, involving the venous system through direct and indirect mechanisms. This paper aims to provide a comprehensive review of these entities, as their awareness can avoid potential diagnostic pitfalls. We also propose a practical classification system of paediatric cerebral venous pathologies, which will enable more accurate reporting of the neuroimaging findings, as relevant to the underlying pathogenesis of these conditions. The proposed classification system comprises of the following main groups: arterio-venous shunting-related disorders, primary venous malformations and veno-occlusive disorders. A multimodal imaging approach has been included in the relevant subsections, with a brief overview of the modality-specific pitfalls that can also limit interpretation of the neuroimaging. The article also summarises the current literature and international practices in terms of management options and outcomes in specific disease entities.