Health Inequity and Time From Pediatric Stroke Onset to Arrival.

BACKGROUND: Time from stroke onset to hospital arrival determines treatment and impacts outcome. Structural, socioeconomic, and environmental factors are associated with health inequity and onset-to-arrival in adult stroke. We aimed to assess the association between health inequity and onset-to-arrival in a pediatric comprehensive stroke center. METHODS: A retrospective observational study was conducted on a consecutive cohort of children (>28 days-18 years) diagnosed with acute arterial ischemic stroke (AIS) between 2004 and 2019. Neighborhood-level material deprivation was derived from residential postal codes and used as a proxy measure for health inequity. Patients were stratified by level of neighborhood-level material deprivation, and onset-to-arrival was categorized into 3 groups: <6, 6 to 24, and >24 hours. Association between neighborhood-level material deprivation and onset-to-arrival was assessed in multivariable ordinal logistic regression analyses adjusting for sociodemographic and clinical factors. RESULTS: Two hundred and twenty-nine children were included (61% male; median age [interquartile range] at stroke diagnosis 5.8-years [1.1-11.3]). Over the 16-year study period, there was an increase in proportion of children diagnosed with AIS living in the most deprived neighborhoods and arriving at the emergency room within 6 hours (P=0.01). Among Asian patients, a higher proportion lived in the most deprived neighborhoods (P=0.02) and level of material deprivation was associated with AIS risk factors (P=0.001). CONCLUSIONS: Our study suggests an increase in pediatric stroke in deprived neighborhoods and certain communities, and earlier arrival times to the emergency room over time. However, whether these changes are due to an increase in incidence of childhood AIS or increased awareness and diagnosis is yet to be determined. The association between AIS risk factors and material deprivation highlights the intersectionality of clinical factors and social determinants of health. Finally, whether material deprivation impacts onset-to-arrival is likely complex and requires further examination.

Brain Tumor Imaging in Adolescents and Young Adults: 2021 WHO Updates for Molecular-based Tumor Types.

Published in 2021, the fifth edition of the World Health Organization (WHO) classification of tumors of the central nervous system (CNS) introduced new molecular criteria for tumor types that commonly occur in either pediatric or adult age groups. Adolescents and young adults (AYAs) are at the intersection of adult and pediatric care, and both pediatric-type and adult-type CNS tumors occur at that age. Mortality rates for AYAs with CNS tumors have increased by 0.6% per year for males and 1% per year for females from 2007 to 2016. To best serve patients, it is crucial that both pediatric and adult radiologists who interpret neuroimages are familiar with the various pediatric- and adult-type brain tumors and their typical imaging morphologic characteristics. Gliomas account for approximately 80% of all malignant CNS tumors in the AYA age group, with the most common types observed being diffuse astrocytic and glioneuronal tumors. Ependymomas and medulloblastomas also occur in the AYA population but are seen less frequently. Importantly, biologic behavior and progression of distinct molecular subgroups of brain tumors differ across ages. This review discusses newly added or revised gliomas in the fifth edition of the CNS WHO classification, as well as other CNS tumor types common in the AYA population.

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

OBJECTIVES: Currently, the BRAF status of pediatric low-grade glioma (pLGG) patients is determined through a biopsy. We established a nomogram to predict BRAF status non-invasively using clinical and radiomic factors. Additionally, we assessed an advanced thresholding method to provide only high-confidence predictions for the molecular subtype. Finally, we tested whether radiomic features provide additional predictive information for this classification task, beyond that which is embedded in the location of the tumor. METHODS: Random forest (RF) models were trained on radiomic and clinical features both separately and together, to evaluate the utility of each feature set. Instead of using the traditional single threshold technique to convert the model outputs to class predictions, we implemented a double threshold mechanism that accounted for uncertainty. Additionally, a linear model was trained and depicted graphically as a nomogram. RESULTS: The combined RF (AUC: 0.925) outperformed the RFs trained on radiomic (AUC: 0.863) or clinical (AUC: 0.889) features alone. The linear model had a comparable AUC (0.916), despite its lower complexity. Traditional thresholding produced an accuracy of 84.5%, while the double threshold approach yielded 92.2% accuracy on the 80.7% of patients with the highest confidence predictions. CONCLUSION: Models that included radiomic features outperformed, underscoring their importance for the prediction of BRAF status. A linear model performed similarly to RF but with the added benefit that it can be visualized as a nomogram, improving the explainability of the model. The double threshold technique was able to identify uncertain predictions, enhancing the clinical utility of the model. CLINICAL RELEVANCE STATEMENT: Radiomic features and tumor location are both predictive of BRAF status in pLGG patients. We show that they contain complementary information and depict the optimal model as a nomogram, which can be used as a non-invasive alternative to biopsy. KEY POINTS: • Radiomic features provide additional predictive information for the determination of the molecular subtype of pediatric low-grade gliomas patients, beyond what is embedded in the location of the tumor, which has an established relationship with genetic status. • An advanced thresholding method can help to distinguish cases where machine learning models have a high chance of being (in)correct, improving the utility of these models. • A simple linear model performs similarly to a more powerful random forest model at classifying the molecular subtype of pediatric low-grade gliomas but has the added benefit that it can be converted into a nomogram, which may facilitate clinical implementation by improving the explainability of the model.

Diagnostic Imaging for Retinoblastoma Cancer Staging: Guide for Providing Essential Insights for Ophthalmologists and Oncologists.

Retinoblastoma is the most common cause of all intraocular pediatric malignancies. It is caused by the loss of RB1 tumor suppressor gene function, although some tumors occur due to MYCN oncogene amplification with normal RB1 genes. Nearly half of all retinoblastomas occur due to a hereditary germline RB1 pathogenic variant, most of which manifest with bilateral tumors. This germline RB1 mutation also predisposes to intracranial midline embryonal tumors. Accurate staging of retinoblastoma is crucial in providing optimal vision-, eye-, and life-saving treatment. The AJCC Cancer Staging Manual has undergone significant changes, resulting in a universally accepted system with a multidisciplinary approach for managing retinoblastoma. The authors discuss the role of MRI and other diagnostic imaging techniques in the pretreatment assessment and staging of retinoblastoma. A thorough overview of the prevailing imaging standards and evidence-based perspectives on the benefits and drawbacks of these techniques is provided. Published under a CC BY 4.0 license. Test Your Knowledge questions for this article are available in the supplemental material.

Imaging of pediatric glioneuronal and neuronal tumors.

Glioneuronal tumors (GNTs) are an expanding group of primary CNS neoplasms, commonly affecting children, adolescents and young adults. Most GNTs are relatively indolent, low-grade, WHO grade I lesions. In the pediatric age group, GNTs have their epicenter in the cerebral cortex and present with seizures. Alterations in the mitogen-activated protein kinase (MAPK) pathway, which regulates cell growth, are implicated in tumorigenesis. Imaging not only plays a key role in the characterization and pre-surgical evaluation of GNTs but is also crucial role in follow-up, especially with the increasing use of targeted inhibitors and immunotherapies. In this chapter, we review the clinical and imaging perspectives of common pediatric GNTs.

Applications of machine learning to MR imaging of pediatric low-grade gliomas.

INTRODUCTION: Machine learning (ML) shows promise for the automation of routine tasks related to the treatment of pediatric low-grade gliomas (pLGG) such as tumor grading, typing, and segmentation. Moreover, it has been shown that ML can identify crucial information from medical images that is otherwise currently unattainable. For example, ML appears to be capable of preoperatively identifying the underlying genetic status of pLGG. METHODS: In this chapter, we reviewed, to the best of our knowledge, all published works that have used ML techniques for the imaging-based evaluation of pLGGs. Additionally, we aimed to provide some context on what it will take to go from the exploratory studies we reviewed to clinically deployed models. RESULTS: Multiple studies have demonstrated that ML can accurately grade, type, and segment and detect the genetic status of pLGGs. We compared the approaches used between the different studies and observed a high degree of variability throughout the methodologies. Standardization and cooperation between the numerous groups working on these approaches will be key to accelerating the clinical deployment of these models. CONCLUSION: The studies reviewed in this chapter detail the potential for ML techniques to transform the treatment of pLGG. However, there are still challenges that need to be overcome prior to clinical deployment.

Accuracy of Information and References Using ChatGPT-3 for Retrieval of Clinical Radiological Information.

Purpose: To assess the accuracy of answers provided by ChatGPT-3 when prompted with questions from the daily routine of radiologists and to evaluate the text response when ChatGPT-3 was prompted to provide references for a given answer. Methods: ChatGPT-3 (San Francisco, OpenAI) is an artificial intelligence chatbot based on a large language model (LLM) that has been designed to generate human-like text. A total of 88 questions were submitted to ChatGPT-3 using textual prompt. These 88 questions were equally dispersed across 8 subspecialty areas of radiology. The responses provided by ChatGPT-3 were assessed for correctness by cross-checking them with peer-reviewed, PubMed-listed references. In addition, the references provided by ChatGPT-3 were evaluated for authenticity. Results: A total of 59 of 88 responses (67%) to radiological questions were correct, while 29 responses (33%) had errors. Out of 343 references provided, only 124 references (36.2%) were available through internet search, while 219 references (63.8%) appeared to be generated by ChatGPT-3. When examining the 124 identified references, only 47 references (37.9%) were considered to provide enough background to correctly answer 24 questions (37.5%). Conclusion: In this pilot study, ChatGPT-3 provided correct responses to questions from the daily clinical routine of radiologists in only about two thirds, while the remainder of responses contained errors. The majority of provided references were not found and only a minority of the provided references contained the correct information to answer the question. Caution is advised when using ChatGPT-3 to retrieve radiological information.

MRI-Based End-To-End Pediatric Low-Grade Glioma Segmentation and Classification.

Purpose: MRI-based radiomics models can predict genetic markers in pediatric low-grade glioma (pLGG). These models usually require tumour segmentation, which is tedious and time consuming if done manually. We propose a deep learning (DL) model to automate tumour segmentation and build an end-to-end radiomics-based pipeline for pLGG classification. Methods: The proposed architecture is a 2-step U-Net based DL network. The first U-Net is trained on downsampled images to locate the tumour. The second U-Net is trained using image patches centred around the located tumour to produce more refined segmentations. The segmented tumour is then fed into a radiomics-based model to predict the genetic marker of the tumour. Results: Our segmentation model achieved a correlation value of over 80% for all volume-related radiomic features and an average Dice score of .795 in test cases. Feeding the auto-segmentation results into a radiomics model resulted in a mean area under the ROC curve (AUC) of .843, with 95% confidence interval (CI) [.78-.906] and .730, with 95% CI [.671-.789] on the test set for 2-class (BRAF V600E mutation BRAF fusion) and 3-class (BRAF V600E mutation BRAF fusion and Other) classification, respectively. This result was comparable to the AUC of .874, 95% CI [.829-.919] and .758, 95% CI [.724-.792] for the radiomics model trained and tested on the manual segmentations in 2-class and 3-class classification scenarios, respectively. Conclusion: The proposed end-to-end pipeline for pLGG segmentation and classification produced results comparable to manual segmentation when it was used for a radiomics-based genetic marker prediction model.

Developmental Curves of the Paediatric Brain Using FLAIR MRI Texture Biomarkers.

Purpose: Analysis of FLAIR MRI sequences is gaining momentum in brain maturation studies, and this study aimed to establish normative developmental curves for FLAIR texture biomarkers in the paediatric brain. Methods: A retrospective, single-centre dataset of 465/512 healthy paediatric FLAIR volumes was used, with one pathological volume for proof-of-concept. Participants were included if the MRI was unremarkable as determined by a neuroradiologist. An automated intensity normalization algorithm was used to standardize FLAIR signal intensity across MRI scanners and individuals. FLAIR texture biomarkers were extracted from grey matter (GM), white matter (WM), deep GM, and cortical GM regions. Sex-specific percentile curves were reported and modelled for each tissue type. Correlations between texture and established biomarkers including intensity volume were examined. Biomarkers from the pathological volume were extracted to demonstrate clinical utility of normative curves. Results: This study analyzed 465 FLAIR sequences in children and adolescents (mean age 10.65 ± 4.22 years, range 2-19 years, 220 males, 245 females). In the WM, texture increased to a maximum at around 8 to 10 years, with different trends between females and males in adolescence. In the GM, texture increased over the age range while demonstrating a local maximum at 8 to 10 years. Texture had an inverse relationship with intensity in the WM across all ages. WM and edema in a pathological brain exhibited abnormal texture values outside of the normative growth curves. Conclusion: Normative curves for texture biomarkers in FLAIR sequences may be used to assess brain maturation and microstructural changes over the paediatric age range.

Impact of cardiopulmonary bypass on cerebrovascular autoregulation assessed by ultrafast ultrasound imaging.

Newborns with congenital heart disease undergoing cardiac surgery are at risk of neurodevelopmental impairment with limited understanding of the impact of intra-operative cardiopulmonary bypass (CPB), deep hypothermia and selective cerebral perfusion on the brain. We hypothesized that a novel ultrasound technique, ultrafast power Doppler (UPD), can assess variations of cerebral blood volume (CBV) in neonates undergoing cardiac surgery requiring CPB. UPD was performed before, during and after surgery in newborns with hypoplastic left heart syndrome undergoing a Norwood operation. We found that global CBV was not significantly different between patients and controls (P = 0.98) and between pre- and post-surgery (P = 0.62). UPD was able to monitor changes in CBV throughout surgery, revealing regional differences in CBV during hypothermia during which CBV correlated with CPB flow rate (R2  = 0.52, P = 0.021). Brain injury on post-operative magnetic resonance imaging was observed in patients with higher maximum variation in CBV. Our findings suggest that UPD can quantify global and regional brain perfusion variation during neonatal cardiac surgery with this first intra-operative application demonstrating an association between CBV and CPB flow rate, suggesting loss of autoregulation. Therefore, the measurement of CBV by UPD could enable optimization of cerebral perfusion during cardiac surgery in neonates. KEY POINTS: The impact of cardiopulmonary bypass (CPB) on the neonatal brain undergoing cardiac surgery is poorly understood. Ultrafast power Doppler (UPD) quantifies cerebral blood volume (CBV), a surrogate of brain perfusion. CBV varies throughout CPB surgery and is associated with variation of the bypass pump flow rate during deep hypothermia. Association between CBV and bypass pump flow rate suggests loss of cerebrovascular autoregulatory processes. Quantitative monitoring of cerebral perfusion by UPD could provide a direct parameter to optimize CPB flow rate.

Magnetic resonance imaging-based scoring of the diseased lung in the preterm infant with bronchopulmonary dysplasia: UNiforme Scoring of the disEAsed Lung in BPD (UNSEAL BPD).

Neonatal chronic lung disease lacks standardized assessment of lung structural changes. We addressed this clinical need by the development of a novel scoring system [UNSEAL BPD (UNiforme Scoring of the disEAsed Lung in BPD)] using T2-weighted single-shot fast-spin-echo sequences from 3 T MRI in very premature infants with and without bronchopulmonary dysplasia (BPD). Quantification of interstitial and airway remodeling, emphysematous changes, and ventilation inhomogeneity was achieved by consensus scoring on a five-point Likert scale. We successfully identified moderate and severe disease by logistic regression [area under the curve (AUC), 0.89] complemented by classification tree analysis revealing gestational age-specific structural changes. We demonstrated substantial interreader reproducibility (weighted Cohen's κ 0.69) and disease specificity (AUC = 0.91). Our novel MRI score enables the standardized assessment of disease-characteristic structural changes in the preterm lung exhibiting significant potential as a quantifiable endpoint in early intervention clinical trials and long-term disease monitoring.

MRI pulmonary artery flow detects lung vascular pathology in preterms with lung disease.

BACKGROUND: Pulmonary vascular disease (PVD) affects the majority of preterm neonates with bronchopulmonary dysplasia (BPD) and significantly determines long-term mortality through undetected progression into pulmonary hypertension. Our objectives were to associate characteristics of pulmonary artery (PA) flow and cardiac function with BPD-associated PVD near term using advanced magnetic resonance imaging (MRI) for improved risk stratification. METHODS: Preterms <32 weeks postmenstrual age (PMA) with/without BPD were clinically monitored including standard echocardiography and prospectively enrolled for 3 T MRI in spontaneous sleep near term (AIRR (Attention to Infants at Respiratory Risks) study). Semi-manual PA flow quantification (phase-contrast MRI; no BPD n=28, mild BPD n=35 and moderate/severe BPD n=25) was complemented by cardiac function assessment (cine MRI). RESULTS: We identified abnormalities in PA flow and cardiac function, i.e. increased net forward volume right/left ratio, decreased mean relative area change and pathological right end-diastolic volume, to sensitively detect BPD-associated PVD while correcting for PMA (leave-one-out area under the curve 0.88, sensitivity 0.80 and specificity 0.81). We linked these changes to increased right ventricular (RV) afterload (RV-arterial coupling (p=0.02), PA mid-systolic notching (t2; p=0.015) and cardiac index (p=1.67×10-8)) and correlated echocardiographic findings. Identified in moderate/severe BPD, we successfully applied the PA flow model in heterogeneous mild BPD cases, demonstrating strong correlation of PVD probability with indicators of BPD severity, i.e. duration of mechanical ventilation (rs=0.63, p=2.20×10-4) and oxygen supplementation (rs=0.60, p=6.00×10-4). CONCLUSIONS: Abnormalities in MRI PA flow and cardiac function exhibit significant, synergistic potential to detect BPD-associated PVD, advancing the possibilities of risk-adapted monitoring.

Time to dismiss boost? Outcomes of children with localized and metastatic germinoma.

PURPOSE: To determine long-term outcomes of a cohort of children with germinoma treated with chemotherapy and radiation therapy without primary tumor boost even in the absence of complete response to chemotherapy METHODS: This retrospective study analyzed the outcome of patients with germinoma consecutively diagnosed and treated at a tertiary care center from January 2000 to December 2021. MRIs were reviewed by two radiologists, blinded to patient data. Tumor location at diagnosis, tumor response to chemotherapy and at completion of radiation therapy and site of relapse were assessed. Tumor response was assessed radiologically by determining the tumor size and response on diffusion-weighted imaging, in addition to biochemical, cytological parameters and neurological status. RESULTS: Of 46 pediatric germinoma patients, 29 children (14 male; median age 12.8 years) received no primary tumor boost. Median follow-up was 63 months (range 9-187 months). Twenty-five children had localized disease and tumor location was suprasellar (n = 11), pineal (n = 10), bifocal (n = 3) and basal ganglia (n = 1) while 4 children had metastatic disease at presentation. All patients completed multi-agent chemotherapy followed by either ventricular irradiation (VI) (23.4 Gy) (n = 23), whole brain (WBI) (23.4 Gy) (n = 5) or craniospinal radiation (CSI) (23.4 Gy) (n = 1). Two children, who had localized disease at presentation and received VI after chemotherapy, relapsed 9 months and 32 months after completion of treatment respectively. No patient had a local relapse. Location of relapse was distant, outside (n = 1) and out- and inside (n = 1) the irradiation field. Five-year progression free survival (PFS) was 91% and overall survival (OS) was 100%. CONCLUSIONS: In this case series, excellent 5-year PFS and OS rates were achieved with chemotherapy followed by radiation therapy of 23.4 Gy delivered without primary tumor boost. No local relapse was observed despite omitting primary tumor boost in patients with localized and metastatic germinoma.

Cochlear Implantation: Systematic Approach to Preoperative Radiologic Evaluation.

Sensorineural hearing loss results from abnormalities that affect the hair cells of the membranous labyrinth, inner ear malformations, and conditions affecting the auditory pathway from the cochlear nerve to the processing centers of the brain. Cochlear implantation is increasingly being performed for hearing rehabilitation owing to expanding indications and a growing number of children and adults with sensorineural hearing loss. An adequate understanding of the temporal bone anatomy and diseases that affect the inner ear is paramount for alerting the operating surgeon about variants and imaging findings that can influence the surgical technique, affect the choice of cochlear implant and electrode type, and help avoid inadvertent complications. In this article, imaging protocols for sensorineural hearing loss and the normal inner ear anatomy are reviewed, with a brief description of cochlear implant devices and surgical techniques. In addition, congenital inner ear malformations and acquired causes of sensorineural hearing loss are discussed, with a focus on imaging findings that may affect surgical planning and outcomes. The anatomic factors and variations that are associated with surgical challenges and may predispose patients to periprocedural complications also are highlighted. © RSNA, 2023 Quiz questions for this article are available through the Online Learning Center. Online supplemental material and the slide presentation from the RSNA Annual Meeting are available for this article.

Bilateral compressive optic neuropathy and outer retinopathy due to optic canal hyperostosis in a child with isolated vitamin a deficiency.

PURPOSE: Vitamin A plays a crucial role in rod phototransduction, with deficient levels manifesting as night blindness. Animal models have demonstrated bone dysplasia in the setting of hypovitaminosis A. We present a rare case of bony overgrowth leading to bilateral compressive optic neuropathy, combined with outer retinopathy, in a paediatric patient secondary to isolated vitamin A deficiency. METHODS: A single case report was conducted from Toronto, Canada. RESULTS: A 12-year-old boy with known autism spectrum disorder presented with a 9-month history of progressive painless vision loss. Vision was 20/300 and hand motion in the right and left eye, respectively. Fundus photography demonstrated bilateral optic atrophy and yellow lesions notably in the right eye far periphery. Optical coherence tomography (OCT) imaging demonstrated thinning of the retinal nerve fibre layer, alterations in the ellipsoid zone, as well as retinal pigment epithelium deposits. Computed tomography imaging demonstrated sphenoid bone thickening with narrow optic canals and moderate optic atrophy bilaterally. Full-field electroretinogram (ERG) demonstrated mildly reduced dark adapted (DA) 0.01 b-wave amplitudes and electronegative configuration of DA 3.0 and DA 10.0 ERG; the light adapted ERGs were normal. The patient was treated with pulse vitamin A therapy. Subsequently, the DA ERG normalized, outer retinal changes reversed and vision stabilised; no surgical intervention was conducted. CONCLUSION: This case represents a rare presentation of compressive optic neuropathy with concomitant outer retinopathy secondary to isolated vitamin A deficiency. Despite improvement in outer retinal integrity on OCT imaging and ERG testing results following vitamin A supplementation, no functional improvement was obtained due to severe optic atrophy.

Variable functional connectivity architecture of the preterm human brain: Impact of developmental cortical expansion and maturation.

Functional connectivity (FC) is known to be individually unique and to reflect cognitive variability. Although FC can serve as a valuable correlate and potential predictor of (patho-) physiological nervous function in high-risk constellations, such as preterm birth, templates for individualized FC analysis are lacking, and knowledge about the capacity of the premature brain to develop FC variability is limited. In a cohort of prospectively recruited, preterm-born infants undergoing magnetic resonance imaging close to term-equivalent age, we show that the overall pattern could be reliably detected with a broad range of interindividual FC variability in regions of higher-order cognitive functions (e.g., association cortices) and less interindividual variability in unimodal regions (e.g., visual and motor cortices). However, when comparing the preterm and adult brains, some brain regions showed a marked shift in variability toward adulthood. This shift toward greater variability was strongest in cognitive networks like the attention and frontoparietal networks and could be partially predicted by developmental cortical expansion. Furthermore, FC variability was reflected by brain tissue characteristics indicating cortical maturation. Brain regions with high functional variability (e.g., the inferior frontal gyrus and temporoparietal junction) displayed lower cortical maturation at birth compared with somatosensory cortices. In conclusion, the overall pattern of interindividual variability in FC is already present preterm; however, some brain regions show increased variability toward adulthood, identifying characteristic patterns, such as in cognitive networks. These changes are related to postnatal cortical expansion and maturation, allowing for environmental and developmental factors to translate into marked individual differences in FC.

Immune signatures of prodromal multiple sclerosis in monozygotic twins.

The tremendous heterogeneity of the human population presents a major obstacle in understanding how autoimmune diseases like multiple sclerosis (MS) contribute to variations in human peripheral immune signatures. To minimize heterogeneity, we made use of a unique cohort of 43 monozygotic twin pairs clinically discordant for MS and searched for disease-related peripheral immune signatures in a systems biology approach covering a broad range of adaptive and innate immune populations on the protein level. Despite disease discordance, the immune signatures of MS-affected and unaffected cotwins were remarkably similar. Twinship alone contributed 56% of the immune variation, whereas MS explained 1 to 2% of the immune variance. Notably, distinct traits in CD4+ effector T cell subsets emerged when we focused on a subgroup of twins with signs of subclinical, prodromal MS in the clinically healthy cotwin. Some of these early-disease immune traits were confirmed in a second independent cohort of untreated early relapsing-remitting MS patients. Early involvement of effector T cell subsets thus points to a key role of T cells in MS disease initiation.

An Educational Graphical User Interface to Construct Convolutional Neural Networks for Teaching Artificial Intelligence in Radiology.

Deep learning techniques using convolutional neural networks (CNNs) have been successfully developed for various medical image analysis tasks. However, the skills to understand and develop deep learning models are not usually taught during radiology training, which constitutes a barrier for radiologists looking to integrate machine learning (ML) into their research or clinical practice. In this work, we developed and evaluated an educational graphical user interface (GUI) to construct CNNs for teaching deep learning concepts to radiology trainees. The GUI was developed in Python using the PyQt and PyTorch frameworks. The functionality of the GUI was demonstrated through a binary classification task on a dataset of MR images of the brain. The usability of the GUI was assessed through 45-min user testing sessions with 5 neuroradiologists and neuroradiology fellows, assessing mean task completion times, the System Usability Scale (SUS), and a qualitative questionnaire as metrics. Task completion times were compared against a ML expert who performed the same tasks. After a 20-min introduction to CNNs and a walkthrough of the GUI, users were able to perform all assigned tasks successfully. There was no significant difference in task completion time compared to a ML expert. The educational GUI achieved a score of 82.5 on the SUS, suggesting that the system is highly usable. Users indicated that the GUI seems useful as an educational tool to teach ML topics to radiology trainees. An educational GUI allows interactive teaching in ML that can be incorporated into radiology training.

Radiomic Features Based on MRI Predict Progression-Free Survival in Pediatric Diffuse Midline Glioma/Diffuse Intrinsic Pontine Glioma.

Purpose: Biopsy-based assessment of H3 K27 M status helps in predicting survival, but biopsy is usually limited to unusual presentations and clinical trials. We aimed to evaluate whether radiomics can serve as prognostic marker to stratify diffuse intrinsic pontine glioma (DIPG) subsets. Methods: In this retrospective study, diagnostic brain MRIs of children with DIPG were analyzed. Radiomic features were extracted from tumor segmentations and data were split into training/testing sets (80:20). A conditional survival forest model was applied to predict progression-free survival (PFS) using training data. The trained model was validated on the test data, and concordances were calculated for PFS. Experiments were repeated 100 times using randomized versions of the respective percentage of the training/test data. Results: A total of 89 patients were identified (48 females, 53.9%). Median age at time of diagnosis was 6.64 years (range: 1-16.9 years) and median PFS was 8 months (range: 1-84 months). Molecular data were available for 26 patients (29.2%) (1 wild type, 3 K27M-H3.1, 22 K27M-H3.3). Radiomic features of FLAIR and nonenhanced T1-weighted sequences were predictive of PFS. The best FLAIR radiomics model yielded a concordance of .87 [95% CI: .86-.88] at 4 months PFS. The best T1-weighted radiomics model yielded a concordance of .82 [95% CI: .8-.84] at 4 months PFS. The best combined FLAIR + T1-weighted radiomics model yielded a concordance of .74 [95% CI: .71-.77] at 3 months PFS. The predominant predictive radiomic feature matrix was gray-level size-zone. Conclusion: MRI-based radiomics may predict progression-free survival in pediatric diffuse midline glioma/diffuse intrinsic pontine glioma.

Loss of tubulin deglutamylase CCP1 causes infantile-onset neurodegeneration.

A set of glutamylases and deglutamylases controls levels of tubulin polyglutamylation, a prominent post-translational modification of neuronal microtubules. Defective tubulin polyglutamylation was first linked to neurodegeneration in the Purkinje cell degeneration (pcd) mouse, which lacks deglutamylase CCP1, displays massive cerebellar atrophy, and accumulates abnormally glutamylated tubulin in degenerating neurons. We found biallelic rare and damaging variants in the gene encoding CCP1 in 13 individuals with infantile-onset neurodegeneration and confirmed the absence of functional CCP1 along with dysregulated tubulin polyglutamylation. The human disease mainly affected the cerebellum, spinal motor neurons, and peripheral nerves. We also demonstrate previously unrecognized peripheral nerve and spinal motor neuron degeneration in pcd mice, which thus recapitulated key features of the human disease. Our findings link human neurodegeneration to tubulin polyglutamylation, entailing this post-translational modification as a potential target for drug development for neurodegenerative disorders.