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.

Neuroanatomy of cerebellar mutism syndrome: the role of lesion location.

Approximately 25% of paediatric patients who undergo cerebellar tumour resection develop cerebellar mutism syndrome. Our group recently showed that damage to the cerebellar deep nuclei and superior cerebellar peduncles, which we refer to as the cerebellar outflow pathway, is associated with an increased risk of cerebellar mutism syndrome. Here, we tested whether these findings replicate in an independent cohort. We evaluated the relationship between lesion location and the development of cerebellar mutism syndrome in an observational study of 56 paediatric patients ranging from five months to 14 years of age who underwent cerebellar tumour resection. We hypothesized that individuals who developed cerebellar mutism syndrome after surgery, relative to those who did not, would have lesions that preferentially intersect with: (i) the cerebellar outflow pathway and (ii) a previously generated 'lesion-symptom map' of cerebellar mutism syndrome. Analyses were conducted in accordance with pre-registered hypotheses and analytic methods (https://osf.io/r8yjv/). We found supporting evidence for both hypotheses. Compared to patients who did not develop cerebellar mutism syndrome, patients with cerebellar mutism syndrome (n = 10) had lesions with greater overlap with the cerebellar outflow pathway (Cohen's d = 0.73, P = 0.05), and the cerebellar mutism syndrome lesion-symptom map (Cohen's d = 1.1, P = 0.004). These results strengthen the association of lesion location with the risk of developing cerebellar mutism syndrome and demonstrate generalizability across cohorts. These findings may help to inform the optimal surgical approach to paediatric cerebellar tumours.

Neuroanatomy of Cerebellar Mutism Syndrome: The Role of Lesion Location.

Approximately 25% of pediatric patients who undergo cerebellar tumor resection develop cerebellar mutism syndrome (CMS). Our group recently showed that damage to the cerebellar deep nuclei and superior cerebellar peduncles, which we refer to as the cerebellar outflow pathway, is associated with increased risk of CMS. Here, we tested whether these findings replicate in an independent cohort. We evaluated the relationship between lesion location and the development of CMS in an observational study of 56 pediatric patients who underwent cerebellar tumor resection. We hypothesized that individuals that developed CMS after surgery (CMS+), relative to those that did not (CMS-) would have lesions that preferentially intersected with: 1) the cerebellar outflow pathway, and 2) a previously generated 'lesion-symptom map' of CMS. Analyses were conducted in accordance with pre-registered hypotheses and analytic methods (https://osf.io/r8yjv/). We found supporting evidence for both hypotheses. Compared with CMS- patients, CMS + patients (n = 10) had lesions with greater overlap with the cerebellar outflow pathway (Cohen's d = .73, p = .05), and the CMS lesion-symptom map (Cohen's d = 1.1, p = .004). These results strengthen the association of lesion location with risk of developing CMS and demonstrate generalizability across cohorts. These findings may help to inform the optimal surgical approach to pediatric cerebellar tumors.

Molecular subgroup of medulloblastoma: evaluation of contribution to CSF diversion following tumour resection.

INTRODUCTION: Medulloblastoma is the commonest malignant brain tumour in children. Pre-operative hydrocephalus is present in up to 90% of these patients at presentation. Following posterior fossa surgery, despite resolution of fourth ventricular obstruction, a proportion of these children will still require cerebrospinal fluid (CSF) diversion for management of persistent or new hydrocephalus. Various scoring systems have been developed to predict the risk for CSF diversion following posterior fossa surgery. However, no accurate tool exists regarding which pathological subset or group of medulloblastoma patients will require a shunt post-operatively. In this study we investigated the impact of molecular subgroup of medulloblastoma on shunt dependency post-operatively in paediatric patients. METHODS: We undertook a retrospective multi-centre study of children with medulloblastoma who underwent tumour resection. Those with available molecular subgroup were identified. Demographic data and clinical parameters including age, sex, presence of pre-operative hydrocephalus, extent of surgical resection, evidence of metastasis/leptomeningeal disease and need for CSF diversion post-operatively were further analysed. RESULTS: Sixty-nine children with medulloblastoma with available molecular data were identified during the study period with male to female ratio of 1.5:1 (42M:27F). Twelve patients (17.4%) belonged to SHH, 10 (14.5%) Wnt, 19 (27.5%) Group 3 and 15 (21.7%) Group 4; 13 (18.8%) were non-specified Group 3 or 4. A total of 18 (26%) patients had evidence of leptomeningeal disease at presentation (20% of Wnt, 42% of Group 3, 33% of group 4, 23% of group 3/4, and 0% of SHH). Fifteen patients (22%) underwent post-operative ventriculoperitoneal (VP) shunt insertion. No patient in the Wnt group required ventriculoperitoneal (VP) shunt post-operatively in this cohort. Need for shunt was associated with pre-operative hydrocephalus, leptomeningeal disease, with molecular group 3 or 4 demonstrating higher rate of leptomeningeal disease, and pre-operative hydrocephalus. Age, extent of resection and pre-operative EVD were not associated with need for shunt in this cohort. Regression analysis identified only pre-operative hydrocephalus and leptomeningeal disease as independent predictors of need for shunt post-resection in this cohort. CONCLUSION: All patients requiring permanent post-operative VP shunt belonged to non-Wnt groups, particularly group 3 and 4. Although medulloblastoma subgroup does not independently predict need for post-operative shunt, presence of leptomeningeal disease and pre-operative hydrocephalus, and their higher prevalence in group 3 and 4, likely account for observed higher rate of shunting in these groups.

Investigating lesion location in relation to cerebellar mutism following pediatric tumor resection.

Background and Objectives: Approximately 25% of pediatric patients who undergo cerebellar tumor resection develop cerebellar mutism syndrome (CMS). Our group recently showed that damage to the cerebellar outflow pathway is associated with increased risk of CMS. Here, we tested whether these findings replicate in an independent cohort. Methods: We evaluated the relationship between lesion location and the development of CMS in an observational study of 56 pediatric patients who underwent cerebellar tumor resection. We hypothesized that individuals that developed CMS after surgery (CMS+), relative to those that did not (CMS-) would have lesions that preferentially intersected with: 1) the cerebellar outflow pathway, and 2) a previously generated 'lesion-symptom map' of CMS. Analyses were conducted in accordance with pre-registered hypotheses and analytic methods (https://osf.io/r8yjv/). Results: We found supporting evidence for both hypotheses. Compared with CMS- patients, CMS+ patients (n=10) had lesions with greater overlap with the cerebellar outflow pathway (Cohen's d=.73, p=.05), and the CMS lesion-symptom map (Cohen's d=1.1, p=.004). Discussion: These results strengthen the association of lesion location with risk of developing CMS and demonstrate generalizability across cohorts. These findings may help to inform the optimal surgical approach to pediatric cerebellar tumors.

Scrotal migration of a ventriculoperitoneal shunt in an adult. A case report and literature review.

Introduction: Scrotal migration of a ventriculoperitoneal shunt (VPS) catheter is a rare complication of VPS. Scrotal migrations usually manifest in the first year post-operatively, usually in the pediatric population, due to processus vaginalis patency and increased abdominal pressure. Research question: To review cases of scrotal migration of a VPS catheter that occur in the adult population, and its recommended management. Material and methods: A case report and review of the literature. Results: A 75-year-old male with a ventriculoperitoneal shunt for normal pressure hydrocephalus, presented with testicular swelling. Imaging revealed that the distal shunt catheter had migrated into his scrotum. He required an emergency shunt revision, with a truncation of the catheter, and involvement of the general surgical team for hernia management. He remained well at one year follow-up. Discussion and conclusion: To the best of our knowledge, this is the fifth case in an adult. This case serves as a reminder to take a thorough clinical history, imaging of the entire VPS pathway, and to consider unusual reasons for VPS failures. Emergency intervention for distal shunt revision is required to prevent further neurological or urological morbidity. Treatment includes not only catheter revision and reinsertion, but the catheter should be truncated, to avoid testicular migration recurrence. Hernia repair can be done either as an emergency or elective case, depending on the patient's clinical status and presentation.

Quality of life reporting in the management of posterior fossa tumours: A systematic review.

Introduction: Survival amongst posterior fossa tumour (PFT) patients is improving. Clinical endpoints such as overall survival fail to depict QoL. There is yet to be a review of current QoL instruments used for adult PFTs. Aim of this review is to outline the QoL reporting in the management of PFTs and measure participation level. Methods: This systematic review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analysis. A search strategy to identify adult patients with PFTs who took part in QoL metrics was conducted. Observational and experimental studies published from 1990 to date were included. Studies with a sample size less than 10 and performance measures such as Karnofsky Performance Status were not considered. Results: A total of 116 studies were included in the final analysis. Vestibular schwannomas were the most common tumour pathology (n = 23,886, 92.6%) followed by pilocytic astrocytomas (n = 657, 2.5%) and meningiomas (n = 437, 1.7%) Twenty-five different QoL measures were used in the study pool. SF-36 was the most common (n = 55, 17 47.4%) QoL metric in the whole study pool, followed by the Penn Acoustic Neuroma QoL scale (n = 24, 20.7%) and Dizziness Handicap Inventory (n = 16, 13.8%). Seventy-two studies reported less-than 100% participation in QoL evaluation. The commonest reason for non-participation was a lack of response (n = 1,718, 60.8%), incomplete questionnaires (n = 268, 9.4%) and cognitive dysfunction (n = 258, 9.1%). Conclusion: Informed clinical decision-making in PFT patients requires the development of specific QoL outcomes. Core outcome sets, and minimal clinically important differences (MCID) are essential for these metrics to show clinically significant improvements in patient QoL.

Left-handedness should not be overrated as a risk factor for postoperative speech impairment in children after posterior fossa tumour surgery: a prospective European multicentre study.

PURPOSE: Cerebellar mutism syndrome (CMS) is a severe neurological complication of posterior fossa tumour surgery in children, and postoperative speech impairment (POSI) is the main component. Left-handedness was previously suggested as a strong risk factor for POSI. The aim of this study was to investigate the relationship between handedness and the risk of POSI. METHODS: We prospectively included children (aged < 18 years) undergoing surgery for posterior fossa tumours in 26 European centres. Handedness was assessed pre-operatively and postoperative speech status was categorised as either POSI (mutism or reduced speech) or habitual speech, based on the postoperative clinical assessment. Logistic regression was used in the risk factor analysis of POSI as a dichotomous outcome. RESULTS: Of the 500 children included, 37 (7%) were excluded from the present analysis due to enrolment at a reoperation; another 213 (43%) due to missing data about surgery (n = 37) and/or handedness (n = 146) and/or postoperative speech status (n = 53). Out of the remaining 250 (50%) patients, 20 (8%) were left-handed and 230 (92%) were right-handed. POSI was observed equally frequently regardless of handedness (5/20 [25%] in left-handed, 61/230 [27%] in right-handed, OR: 1.08 [95% CI: 0.40-3.44], p = 0.882), also when adjusted for tumour histology, location and age. CONCLUSION: We found no difference in the risk of POSI associated with handedness. Our data do not support the hypothesis that handedness should be of clinical relevance in the risk assessment of CMS.

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.

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.

Improved prediction of postoperative pediatric cerebellar mutism syndrome using an artificial neural network.

BACKGROUND: Postoperative pediatric cerebellar mutism syndrome (pCMS) is a common but severe complication that may arise following the resection of posterior fossa tumors in children. Two previous studies have aimed to preoperatively predict pCMS, with varying results. In this work, we examine the generalization of these models and determine if pCMS can be predicted more accurately using an artificial neural network (ANN). METHODS: An overview of reviews was performed to identify risk factors for pCMS, and a retrospective dataset was collected as per these defined risk factors from children undergoing resection of primary posterior fossa tumors. The ANN was trained on this dataset and its performance was evaluated in comparison to logistic regression and other predictive indices via analysis of receiver operator characteristic curves. The area under the curve (AUC) and accuracy were calculated and compared using a Wilcoxon signed-rank test, with P < .05 considered statistically significant. RESULTS: Two hundred and four children were included, of whom 80 developed pCMS. The performance of the ANN (AUC 0.949; accuracy 90.9%) exceeded that of logistic regression (P < .05) and both external models (P < .001). CONCLUSION: Using an ANN, we show improved prediction of pCMS in comparison to previous models and conventional methods.

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.

Postoperative hydrocephalus management may cause delays in adjuvant treatment following paediatric posterior fossa tumour resection: a multicentre retrospective observational study.

INTRODUCTION: Hydrocephalus persists in 10-40% of children with posterior fossa tumours (PFT). A delay in commencement of adjuvant therapy (AT) can negatively influence survival. The objective of this study was to determine whether postoperative cerebrospinal fluid (CSF) diversion procedures caused potentially preventable delays in AT. METHODS: A retrospective study of children diagnosed with PFT requiring AT from 2004 to 2018 from two large centres was conducted. Data on histology, timing of ventriculo-peritoneal shunt (VPS) insertion, and AT was collected. The modified Canadian Preoperative Prediction Rule for Hydrocephalus (mCPPRH) score was calculated. The primary outcome was delay in AT beyond 40 days post-resection. Progression-free and overall survival were assessed. RESULTS: Out of 196 primary PFT resections, 144 fitted the inclusion criteria. Mean age was 6.57 ± 4.62. Histology was medulloblastoma (104), ependymoma (27), and others (13). Forty patients had a VPS inserted; 17 of these experienced a delay in AT. A total of 104 patients were not shunted; 15 of these had delayed AT (p = 0.0007). Patients who had a VPS insertion had longer intervals from surgery to commencement of AT (34.5 vs 30.8, p = 0.05). There was no significant difference in mCPPRH score between those who had a VPS (4.03) and those who did not (3.61; p = 0.252). Multivariable linear regression modelling did not show a significant effect of VPS or mCPPRH on progression-free survival or OS. CONCLUSION: CSF diversion procedures may cause a preventable delay in the initiation of adjuvant therapy. Early post-operative VP shunt insertion, rather than a 'wait and see policy' should be considered in order to reduce this delay.

Postoperative speech impairment and surgical approach to posterior fossa tumours in children: a prospective European multicentre cohort study.

BACKGROUND: Brain tumours are the most common solid tumours in childhood. Half of these tumours occur in the posterior fossa, where surgical removal is complicated by the risk of cerebellar mutism syndrome, of which postoperative speech impairment (POSI) is a cardinal symptom, in up to 25% of patients. The surgical approach to midline tumours, mostly undertaken by transvermian or telovelar routes, has been proposed to influence the risk of POSI. We aimed to investigate the risk of developing POSI, the time course of its resolution, and its association with surgical approach and other clinical factors. METHODS: In this observational prospective multicentre cohort study, we included children (aged <18 years) undergoing primary surgery for a posterior fossa tumour at 26 centres in nine European countries. Within 72 h of surgery, the operating neurosurgeon reported details on the tumour location, surgical approach used, duration of surgery, use of traction, and other predetermined factors, using a standardised surgical report form. At 2 weeks, 2 months, and 1 year after surgery, a follow-up questionnaire was filled out by a paediatrician or neurosurgeon, including neurological examination and assessment of speech. Speech was classified as mutism, reduced speech, or habitual speech. POSI was defined as either mutism or severely reduced speech. Ordinal logistic regression was used to analyse the risk of POSI. FINDINGS: Between Aug 11, 2014, and Aug 24, 2020, we recruited 500 children. 426 (85%) patients underwent primary tumour surgery and had data available for further analysis. 192 (45%) patients were female, 234 (55%) patients were male, 81 (19%) patients were aged 0-2 years, 129 (30%) were aged 3-6 years, and 216 (51%) were aged 7-17 years. 0f 376 with known postoperative speech status, 112 (30%) developed POSI, 53 (14%) developed mutism (median 1 day [IQR 0-2]; range 0-10 days), and 59 (16%) developed reduced speech after surgery (0 days [0-1]; 0-4 days). Mutually adjusted analyses indicated that the independent risk factors for development of POSI were younger age (linear spline, p=0·0087), tumour location (four levels, p=0·0010), and tumour histology (five levels, p=0·0030); surgical approach (six levels) was not a significant risk factor (p=0·091). Tumour location outside the fourth ventricle and brainstem had a lower risk of POSI (with fourth ventricle as reference, odds ratio (OR) for cerebellar vermis 0·34 [95% CI 0·14-0·77] and OR for cerebellar hemispheres 0·23 [0·07-0·70]). Compared with pilocytic or pilomyxoid astrocytoma, a higher risk of POSI was seen for medulloblastoma (OR 2·85 [1·47-5·60]) and atypical teratoid rhabdoid tumour (10·30 [2·10-54·45]). We did not find an increased risk of POSI for transvermian surgical approach compared with telovelar (0·89 [0·46-1·73]). Probability of speech improvement from mutism reached 50% around 16 days after mutism onset. INTERPRETATION: Our data suggest that a midline tumour location, younger age, and high-grade tumour histology all increase the risk of speech impairment after posterior fossa tumour surgery. We found no evidence to recommend a preference for telovelar over transvermian surgical approach in the management of posterior fossa tumours in children in relation to the risk of developing POSI. FUNDING: The Danish Childhood Cancer Foundation, the Swedish Childhood Cancer Foundation, the UK Brain Tumour Charity, the Danish Cancer Society, Det Kgl Kjøbenhavnske Skydeselskab og Danske Broderskab, the Danish Capitol Regions Research Fund, Dagmar Marshall Foundation, Rigshospitalet's Research Fund, and Brainstrust.

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.

Tractographic and Microstructural Analysis of the Dentato-Rubro-Thalamo-Cortical Tracts in Children Using Diffusion MRI.

The dentato-rubro-thalamo-cortical tract (DRTC) is the main outflow pathway of the cerebellum, contributing to a finely balanced corticocerebellar loop involved in cognitive and sensorimotor functions. Damage to the DRTC has been implicated in cerebellar mutism syndrome seen in up to 25% of children after cerebellar tumor resection. Multi-shell diffusion MRI (dMRI) combined with quantitative constrained spherical deconvolution tractography and multi-compartment spherical mean technique modeling was used to explore the frontocerebellar connections and microstructural signature of the DRTC in 30 healthy children. The highest density of DRTC connections were to the precentral (M1) and superior frontal gyri (F1), and from cerebellar lobules I-IV and IX. The first evidence of a topographic organization of anterograde projections to the frontal cortex at the level of the superior cerebellar peduncle (SCP) is demonstrated, with streamlines terminating in F1 lying dorsomedially in the SCP compared to those terminating in M1. The orientation dispersion entropy of DRTC regions appears to exhibit greater contrast than that shown by fractional anisotropy. Analysis of a separate reproducibility cohort demonstrates good consistency in the dMRI metrics described. These novel anatomical insights into this well-studied pathway may prove to be of clinical relevance in the surgical resection of cerebellar tumors.

Surveillance imaging of grade 1 astrocytomas in children: can duration and frequency of follow-up imaging and the use of contrast agents be reduced?

PURPOSE: The optimum strategy for the surveillance of low-grade gliomas in children has not been established, and there is concern about the use of gadolinium-based contrast agents (GBCAs), particularly in children, due to their deposition in the brain. The number of surveillance scans and the use of GBCAs in surveillance of low-risk tumours should ideally be limited. We aimed to investigate the consistency and utility of our surveillance imaging and also determine to what extent the use of GBCAs contributed to decisions to escalate treatment in children with grade 1 astrocytomas. METHODS: This was a retrospective single-centre study at a tertiary paediatric hospital. All children with a new diagnosis of a non-syndromic World Health Organization (WHO) grade 1 astrocytoma between 2007 and 2013 were included, with surveillance imaging up to December 2018 included in analysis. The intervals of surveillance imaging were recorded, and imaging and electronic health records were examined for decisions related to treatment escalation. RESULTS: Eighty-eight patients had 690 surveillance scans in the study period. Thirty-one patients had recurrence or progression leading to treatment escalation, 30 of whom were identified on surveillance imaging. The use of GBCAs did not appear to contribute to multidisciplinary team (MDT) decisions in the majority of cases. CONCLUSION: Surveillance imaging could be reduced in number and duration for completely resected cerebellar tumours. MDT decisions were rarely made on the basis of post-contrast imaging, and GBCA administration could therefore potentially be restricted in the setting of surveillance of grade 1 astrocytomas in children.

Neurosurgical applications of tractography in the UK.

INTRODUCTION: Tractography derived from diffusion MRI can provide important insights into human brain microstructure in vivo. Neurosurgeons were quick to adopt the technique at the turn of the century, but it remains plagued by technical fallibilities. This study aims to describe how tractography is deployed clinically in a modern-day, public healthcare system, serving as a snapshot from the 'shop floor' of British neurosurgical practice. METHODS: An 11-question survey was circulated to the mailing lists of the Society of British Neurological Surgeons and British Neurosurgical Trainees' Association, including questions on frequency, indication, tracts reconstructed, specific details of techniques used and personnel by whom it was performed, and a free-text section on the limitations of tractography. RESULTS: 58 survey responses were received, covering all 40 neurosurgical units in the UK and Ireland. Overall, responses were received from neurosurgeons at 36 units (90.0%) stating tractography was in use at that unit. 74.1% of the responses were from Consultants. The most common indication for tractography was in tumour resection. It was most commonly performed by neuroradiologists or imaging scientists. 75.9% of respondents stated that the model used to process tractography was the diffusion tensor (DTI). Many respondents were unaware of which algorithm (74.1%) or software tools (65.6%) were used by the operator to produce tractography visualisations. The corticospinal tract was the most commonly reconstructed tract. The most commonly cited limitations of the technique were perceived inaccuracy and brain shift. CONCLUSIONS: In this UK-based survey of practising neurosurgeons, we show that 90% of neurosurgical units in the UK and Ireland use tractography regularly; that predominantly DTI-based reconstructions are used; that tumour resection remains the most frequent use of the technique; and that large tracts such as the corticospinal tract are most frequently identified. Many neurosurgeons remain unfamiliar with the underlying methods used to produce tractography visualisations.

Fourth ventricle tumors in children: complications and influence of surgical approach.

OBJECTIVES: The goal of this study was to characterize the complications and morbidity related to the surgical management of pediatric fourth ventricle tumors. METHODS: All patients referred to the authors' institution with posterior fossa tumors from 2002 to 2018 inclusive were screened to include only true fourth ventricle tumors. Preoperative imaging and clinical notes were reviewed to extract data on presenting symptoms; surgical episodes, techniques, and adjuncts; tumor histology; and postoperative complications. RESULTS: Three hundred fifty-four children with posterior fossa tumors were treated during the study period; of these, 185 tumors were in the fourth ventricle, and 167 fourth ventricle tumors with full data sets were included in this analysis. One hundred patients were male (mean age ± SD, 5.98 ± 4.12 years). The most common presenting symptom was vomiting (63.5%). The most common tumor types, in order, were medulloblastoma (94 cases) > pilocytic astrocytoma (30 cases) > ependymoma (30 cases) > choroid plexus neoplasms (5 cases) > atypical teratoid/rhabdoid tumor (4 cases), with 4 miscellaneous lesions. Of the 67.1% of patients who presented with hydrocephalus, 45.5% had an external ventricular drain inserted (66.7% of these prior to tumor surgery, 56.9% frontal); these patients were more likely to undergo ventriculoperitoneal shunt (VPS) placement at a later date (p = 0.00673). Twenty-two had an endoscopic third ventriculostomy, of whom 8 later underwent VPS placement. Overall, 19.7% of patients had a VPS sited during treatment.Across the whole series, the transvermian approach was more frequent than the telovelar approach (64.1% vs 33.0%); however, the telovelar approach was significantly more common in the latter half of the series (p < 0.001). Gross-total resection was achieved in 70.7%. The most common postoperative deficit was cerebellar mutism syndrome (CMS; 28.7%), followed by new weakness (24.0%), cranial neuropathy (18.0%), and new gait abnormality/ataxia (12.6%). Use of intraoperative ultrasonography significantly reduced the incidence of CMS (p = 0.0365). There was no significant difference in the rate of CMS between telovelar or transvermian approaches (p = 0.745), and multivariate logistic regression modeling did not reveal any statistically significant relationships between CMS and surgical approach. CONCLUSIONS: Surgical management of pediatric fourth ventricle tumors continues to evolve, and resection is increasingly performed through the telovelar route. CMS is enduringly the major postoperative complication in this patient population.