Atypical teratoid rhabdoid tumor (AT/RT) is a rare, highly aggressive embryonal central nervous system malignancy occurring predominately in infants and toddlers. Spinal AT/RT (spAT/RT) cases are even more limited, and as a result, little is known regarding prognostic factors and optimal treatment regimens. Molecularly, AT/RT is divided into three groups: AT/RT-SHH, AT/RT-TYR and AT/RT-MYC. spAT/RT is predominantly of the MYC subtype. Additionally, a third of patients with AT/RT have a germline Rhabdoid Tumor Predisposition Syndrome (RTPS) that increases the likelihood of developing additional rhabdoid tumors, including renal rhabdoid tumors. Due to the rarity of these tumors, there is a lack of consensus on treatment strategies to be employed. This review paper details the published literature on spAT/RT, with particular emphasis on the recent advances in understanding the biology of these aggressive tumors and currently available therapeutic options, and highlights the challenges associated with the management of this extremely rare condition.

Summary What is already known about this topic? What does this study add?

To develop and evaluate deep learning methods for synthesizing 3D brain metastases (BM) on magnetic resonance (MR) images to improve downstream BM detection and segmentation performances for robust clinical detection and streamlined treatment-planning workflows, T1-weighted MR images of 1832 patients with 10,276 BM were divided into training (80%), validation (10%), and test (10%) datasets for training the BM synthesis models and downstream models for BM detection and segmentation. A 3D-2D generative adversarial network with controllable configurations for 3D BM synthesis was proposed. The network consists of two 3D generators to create 3D lesion intermediate representations controlling the lesion’s characteristics and 3D continuity, and a 2D generator with a 2D perceptual loss to generate realistic lesion images slice by slice. Training with synthetic data (SD) and conventional augmentation (CA) on the downstream tasks were compared using two-sided pairwise Wilcoxon signed rank test. SD showed improvement on the BM detection and segmentation downstream tasks with limited training data. Especially, when 10% of the original training data was available, adding 176 SD improved BM segmentation Dice by 2.9% relative to CA alone (0.665 vs. 0.646, p < 0.0001). With 8000 SD, segmentation boundary accuracy further improved, reducing HD95 by 10.5% and ASSD by 23.5% compared with CA alone.

RATIONALE AND OBJECTIVES:The purpose of this study is to evaluate whether the whole-body 0.55 T MRI may be a clinically acceptable or non-inferior alternative to 1.5T/3T for internal auditory canal (IAC) imaging due to greater field homogeneity and resultant reduced susceptibility artifacts near air-bone interfaces. We compare image quality, visibility of IAC anatomy, and cochlear measurements at 0.55 T vs. 1.5T/3T. MATERIALS AND METHODS:In this IRB-approved retrospective study, 25 patients who underwent 0.55 T MRI using IAC protocol with prior/ subsequent comparative scans at 1.5T/3T scans were selected. A subset (n=13) also had high-resolution CT (HRCT). 3 board-certified neuroradiologists independently scored the visibility of anatomical structures and overall image quality on a 4-point Likert scale in one non-pathological ear. Readers also measured cochlear length, cochlear width, and IAC width on double oblique coronal images. RESULTS:All three readers rated the visualization of anatomic structures as clinically acceptable (scores ≥ 2) at 0.55 T. Two of the three readers reported no significant difference in visualization of these structures between 0.55 T and higher field strengths. Quantitative measurements (cochlear length, width, and IAC width) at 0.55 T were comparable to 1.5T/3T. One reader (R2) rated overall image quality significantly lower at 0.55 T (mean score: 3.1 at 0.55 T vs. 2.6 at HF; p = 0.005). CONCLUSION:Clinical evaluation of IAC and inner ear structures using a 3D T2w SPACE sequence at 0.55 T is an acceptable alternative to imaging at conventional field strengths of 1.5/3T.

Perineural spread (PNS) is a clinically significant route of tumor extension in head and neck malignancies and is not uncommonly overlooked on imaging. Often clinically occult, PNS may first be suggested by radiologists, and its detection can have significant implications for prognosis and management. This article provides a comprehensive overview of PNS, including relevant cranial nerve anatomy, optimized imaging protocols, and characteristic radiologic features. Multiple illustrative cases are included to improve recognition and highlight the critical role of imaging in the early identification of this often occult disease process.

The analysis of cell-free tumor DNA (ctDNA) and proteins in the blood of patients with cancer potentiates a new generation of non-invasive diagnostic approaches. However, confident detection of tumor-originating markers is challenging, especially in the context of brain tumors, where these analytes in plasma are extremely scarce. Here, we apply a sensitive single-molecule technology to profile multiple histone modifications on individual nucleosomes from the plasma of patients with diffuse midline glioma (DMG). The system reveals epigenetic patterns unique to DMG, significantly differentiating this group of patients from healthy subjects or individuals diagnosed with other cancer types. We further develop a method to directly quantify the tumor-originating oncoproteins, lysine 27 to methionine substitution in histone H3 (H3-K27M) and mutant p53, from <1 mL of plasma, allowing for the accurate molecular classification of patients with DMG. We show that our strategy correlates with MRI and droplet-digital PCR (ddPCR) measurements of ctDNA, highlighting the clinical potential of single-molecule-based, multi-parametric assays for DMG diagnosis and treatment monitoring.

Abstract ONC201 is the first monotherapy to improve outcomes in H3K27M- diffuse midline glioma (DMG) beyond radiation. Despite impressive early efficacy in H3K27M-DMG, individual response to ONC201 is variable and no radiographic biomarkers predict long term response. Previous studies demonstrated baseline MRI diffusion apparent diffusion coefficient (ADC) is lower in DMG with H3K27M compared to wildtype but change after radiation therapy was not correlated with improved OS or PFS. We hypothesize ADC will enable stratification of patients more likely to respond to ONC201. Imaging and clinical data were abstracted from chart review of patients treated with ONC201 at University of Michigan. Two neuroradiologists performed centralized review for RAPNO measurements and mean ADC was assessed using ROI measured in consistent anatomic locations, with areas of cystic degeneration or necrosis excluded. Sixty-one patients were identified for review. Preliminary analysis was performed on twenty-three patients, with upfront ONC201 therapy, median age 14.8 years old (5-25yo), primary site includes pontine (n=15), thalamic (n=7). Baseline characteristics of age, ADC and size at diagnosis or follow-up RAPNO scored size were not statistically significant. Patients with longer than median survival (334 days) demonstrated increased ADC (+58.25x10-6mm2/s) from baseline to cycle 2 whereas patients with shorter than median survival demonstrated decreased ADC (-174.8x10-6mm2/s) (p = 0.0076). Further, patients with decreased ADC demonstrated mean OS of 302.8 days and patients with increased ADC had mean OS of 588.0 (p = 0.0054). In H3K27M-DMG patients treated with ONC201, increased ADC after two cycles of ONC201 was strongly predictive of longer overall survival. Our recent work demonstrated that ONC201 disrupts metabolic and epigenetic pathways to restore pathognomonic H3K27me3 reduction, which may underline greater change in ADC. Ongoing work will validate these findings in an independent external cohort and integrate histogram analysis, parametric assessments, MRI perfusion and liquid biopsy biomarkers (cf-tDNA and metabolomics).

PURPOSE:The purpose of this study was to investigate an extended self-adapting nnU-Net framework for detecting and segmenting brain metastases (BM) on magnetic resonance imaging (MRI). METHODS AND MATERIALS:Six different nnU-Net systems with adaptive data sampling, adaptive Dice loss, or different patch/batch sizes were trained and tested for detecting and segmenting intraparenchymal BM with a size ≥2 mm on 3 Dimensional (3D) post-Gd T1-weighted MRI volumes using 2092 patients from 7 institutions (1712, 195, and 185 patients for training, validation, and testing, respectively). Gross tumor volumes of BM delineated by physicians for stereotactic radiosurgery were collected retrospectively and curated at each institute. Additional centralized data curation was carried out to create gross tumor volumes of uncontoured BM by 2 radiologists to improve the accuracy of ground truth. The training data set was augmented with synthetic BMs of 1025 MRI volumes using a 3D generative pipeline. BM detection was evaluated by lesion-level sensitivity and false-positive (FP) rate. BM segmentation was assessed by lesion-level Dice similarity coefficient, 95-percentile Hausdorff distance, and average Hausdorff distance (HD). The performances were assessed across different BM sizes. Additional testing was performed using a second data set of 206 patients. RESULTS:Of the 6 nnU-Net systems, the nnU-Net with adaptive Dice loss achieved the best detection and segmentation performance on the first testing data set. At an FP rate of 0.65 ± 1.17, overall sensitivity was 0.904 for all sizes of BM, 0.966 for BM ≥0.1 cm3, and 0.824 for BM <0.1 cm3. Mean values of Dice similarity coefficient, 95-percentile Hausdorff distance, and average HD of all detected BMs were 0.758, 1.45, and 0.23 mm, respectively. Performances on the second testing data set achieved a sensitivity of 0.907 at an FP rate of 0.57 ± 0.85 for all BM sizes, and an average HD of 0.33 mm for all detected BM. CONCLUSIONS:Our proposed extension of the self-configuring nnU-Net framework substantially improved small BM detection sensitivity while maintaining a controlled FP rate. Clinical utility of the extended nnU-Net model for assisting early BM detection and stereotactic radiosurgery planning will be investigated.

Thyroid cancer is one of the most rapidly increasing malignancies in the Western world. The most common forms of thyroid cancer include papillary, follicular, Hurthle cell, anaplastic, and medullary thyroid cancer. The patterns of spread, imaging characteristics, staging, and treatment of these cancers vary based on subtype and tumor characteristics. Ultrasound imaging, radionuclide/positron emission tomography, computed tomography, and magnetic resonance imaging may all be used in the evaluation, staging, and posttherapeutic monitoring of thyroid cancer. Surgical resection and radioiodine ablation therapy are considered the cornerstone treatments for the majority of thyroid cancers. Following treatment, biochemical monitoring, imaging follow-up, and clinical assessment constitute the mainstays of thyroid cancer surveillance.

Recent technological advances in deep learning (DL) have led to more accurate brain metastasis (BM) detection. As a data driven approach, DL’s performance highly relies on the size and quality of the training data. However, collecting large amount of medical data is costly, and it’s difficult to include BMs with various locations, sizes, and structures etc. Thus, we propose a 3D-2D GAN for fully 3D BM synthesis with configurable parameters. First, two 3D networks are used to synthesize the mask and quantized intensity map of a lesion from 3 concentric spheres, which are used to control the lesion’s location, size and structure. Then, a 2D network is used to synthesize the final lesion with proper appearance from the quantized intensity map and the background MR image. With this 3D-2D design, the 3D networks enable the synthetic metastasis to be spatially continuous in all 3 dimensions through the guidance of the 3D intermediate presentation of the lesion, while the 2D network enables the use of 2D perceptual loss to make the final synthesized lesion look realistic. In addition, different network up-sampling strategies and postprocessing are used to control the heterogeneity and contrast of the synthetic lesion. All the synthesized images were reviewed by a radiologist. The indistinguishability rate of the synthesized lesion is above 70%. The configurable parameters for the lesion’s location, size, and structure, heterogeneity and contrast were reviewed to be effective. Our work demonstrates the feasibility of synthesizing configurable 3D BM lesions for fully 3D data augmentation.

Brain metastases are the most common malignant form of tumors and occur in 10%-30% of adult patients with systematic cancer. With recent advances in treatment options, there is an increasing evidence that automated detection and segmentation from MRI can assist clinicians for diagnosis and therapy planning. In this study, we investigate the impact of data domain on self-supervised learning (SSL) for pretraining a deep learning network to detect and segment brain metastases on 3D post-contrast T1-weighted images. We performed pretraining a 3D patch-based U-Net using the Model Genesis framework on three subject cohorts that have different data domain. The pretrained networks were then finetuned on brain MR scans from patients with metastases as a downstream task dataset. We analyzed the impact of data domain on SSL by examining validation metric evolution, FROC analyses and testing performance of early-trained models and best-validated models. Our results suggested that, in the early stage of finetuning for the target task, SSL is crucial for faster training convergence and similar data domain on SSL could be helpful to attain improved detection and segmentation performance earlier. However, we observed that the importance of data domain similarity for SSL progressively diminished as training continued with sufficient amount of iterations in our relatively large data regime. After training convergence, the best-validated models pretrained with SSL provided enhanced detection performance over the model without pretraining regardless of data domain.

Amyloidomas are focal solitary amyloid masses without systemic involvement that have been observed to occur in various body locations. When presenting intracranially, they pose a challenging diagnostic and therapeutic course given their location and rarity. We report a case of a 62-year-old man with a 4-year history of seizure and headaches. Magnetic resonance imaging was initially inconclusive but revealed an ill-defined right temporal lobe lesion. Biopsy later confirmed a cerebral amyloidoma. We also review the current literature on the pathogenesis, imaging findings, prognosis, and treatment of cerebral amyloidomas.

INTRODUCTION: Benign External Hydrocephalus (BEH) is a common condition seen by pediatric neurologists and neurosurgeons in macrocephalic infants. It is accepted that patients with BEH are at increased risk of spontaneous subdural hematoma (SDHs) formation. METHODS: A large retrospective chart review was performed at a single institution for patients 2 years or younger with a diagnosis of BEH by neurology or neurosurgery and head circumference >85th percentile. Demographic data, head circumference, and occurrence of subdural hematoma were extracted. MRIs were reviewed by two neuroradiologists and the SAS size measured. SAS size was graded per the Tucker et al grading system (<5 mm width is grade 0, 5-9 mm is grade 1 and > 10 mm is grade 2) for consistency. RESULTS: Over 2.5 million patients' charts were queried from a single institution and 480 patients met the inclusion criteria. Of that population, 187 children had an MRI of the brain available for review. The prevalence of spontaneous SDH in this BEH population over 25 years was 8.12% (39 patients of 480). The average size of the SAS was 5.86 mm (± 2.29 mm). A majority, 64.7%, had grade 1 SAS. About 39% had ventriculomegaly. There was no significant association between SAS grade and prevalence of SDHs (p = 0.124). SAS size was not significantly different between those with and without SDH (p= 0.402). CONCLUSION: BEH is a common condition seen by pediatric neurologists and neurosurgeons. Infants with BEH are thought to have increased risk of spontaneous SDH formation compared to the general population. A majority of infants had grade 1 SAS. The grade of SAS was not significantly associated with SDH formation and development of SDH was not associated with larger SAS size.

PURPOSE:To construct, apply, and evaluate a multidisciplinary approach in teaching radiology to Canadian medical students.METHODS:A multidisciplinary team of radiology and other disciplines experts designed an online 5-session course that was delivered to medical students. The topics of each session were clinical cases involving different systems. The target audience was medical students of Canadian schools. Pretests and post-tests were administered before and after each session respectively. An evaluation survey was distributed at the end of the course to gauge students' perceptions of this experience.RESULTS:An average of 425 medical students attended the live sessions. For each session, 405 students completed both the pre-tests and post-tests. In general, students scored an average of 56% higher on the post-test than on the pre-test. The final course survey was completed by 469 students. The survey results show that more than 98% of students found the course to meet or exceed their expectations. Over 80% of students agreed that the course increased their interest in radiology and about 81% agree that the topics presented were excellent and clinically important. The ratings in the final survey results also indicate that students increased their confidence in basic radiology skills after completing the course.CONCLUSIONS:The implementation of an integrative clinical approach to teaching radiology in a virtual setting is achievable. It provides efficient use of educational resources while being accessible by a large number of students across different medical schools.

OBJECTIVE Benign expansion of the subarachnoid spaces (BESS) is a condition seen in macrocephalic infants. BESS is associated with mild developmental delays which tend to resolve within a few years. It is accepted that patients with BESS are at increased risk of spontaneous subdural hematomas (SDHs), although the exact pathophysiology is not well understood. The prevalence of spontaneous SDH in BESS patients is poorly defined, with only a few large single-center series published. In this study the authors aimed to better define BESS prevalence and developmental outcomes through the longitudinal review of a large cohort of BESS patients. METHODS A large retrospective review was performed at a single institution from 1995 to 2020 for patients 2 years of age or younger with a diagnosis of BESS by neurology or neurosurgery and head circumference > 85th percentile. Demographic data, head circumference, presence of developmental delay, occurrence of SDH, and need for surgery were extracted from patient charts. The subarachnoid space (SAS) size was measured from the available MR images, and the sizes of those who did and did not develop SDH were compared. RESULTS Free text search revealed BESS mentioned within the medical records of 1410 of 2.6 million patients. After exclusion criteria, 480 patients remained eligible for the study. Thirty-two percent (n = 154) of patients were diagnosed with developmental delay, most commonly gross motor delay (53%). Gross motor delay resolved in 86% of patients at a mean age of 22.2 months. The prevalence of spontaneous SDH in this BESS population over a period of 25 years was 8.1%. There was no significant association between SAS size and SDH formation. CONCLUSIONS This study represents results for one of the largest cohorts of patients with BESS at a single institution. Gross motor delay was the most common developmental delay diagnosed, and a majority of patients had resolution of their delay. These data support that children with BESS have a higher prevalence of SDH than the general pediatric population, although SAS size was not significantly associated with SDH development.

Purpose: To assess associations between imaging biomarkers from standard of care pre-treatment CT and FDG-PET scans and locoregional (LR) and distant metastatic (DM) recurrences in patients with p16+ oropharyngeal squamous cell carcinoma (OPSCC) treated with definitive chemoradiotherapy (CRT). Methods: An institutional database from a single NCI-designated cancer center identified 266 patients with p16+ OPSCC treated with definitive CRT in our department from 2005 to 2016 with evaluable pre-treatment FDG-PET scans. Quantitative SUV metrics and qualitative imaging metrics were determined from FDG-PET and CT scans, while clinical characteristics were abstracted from the medical record. Associations between clinical/imaging features and time to LR (TTLRF) or DM (TTDMF) failure and overall survival (OS) were assessed using univariable Cox regression and penalized stepwise regression for multivariable analyses (MVA). Results: There were 27 LR and 32 DM recurrences as incident failures. Imaging biomarkers were significantly associated with TTLRF, TTDMF and OS. FDG-PET metrics outperformed CT and clinical metrics for TTLRF, with metabolic tumor volume being the only significant feature selected on MVA: Cindex = 0.68 (p = 0.01). Radiographic extranodal extension (rENE), positive retropharyngeal nodes (RPN +), and clinical stage were significant on MVA for TTDMF: C-index = 0.84 (p < 0.001). rENE, group stage, and RPN+ were significant on MVA for OS: C-index = 0.77 (p < 0.001). Conclusions: In the largest study to date of uniformly treated patients with CRT to evaluate both pretreatment CT and FDG-PET, radiographic biomarkers were significantly associated with TTLRF, TTDMF and OS among patients with p16+ OPSCC treated with CRT. CT metrics performed best to predict TTDMF, while FDG-PET metrics showed improved prediction for LRRFS. These metrics may help identify candidates for treatment intensification or de-escalation of therapy. Statement of translational relevance: Pre-treatment imaging features from standard-of-care PET/CT imaging show promise for predicting long-term outcomes following HPV-associated oropharynx cancer (HPVOPC) therapy. This study comprehensively characterizes qualitative and quantitative pre-treatment imaging metrics associated with time to pattern-specific failure in a cohort of 266 patients treated uniformly with definitive chemoradiation. Multivariate analysis (MVA) for time to locoregional failure (TTLRF), time to distant metastatic failure (TTDMF), and overall survival (OS) was performed. FDG-PET metrics outperformed CT and clinical metrics for TTLRF. CT radiographic extranodal extension, positive retropharyngeal nodes, and stage strongly predicted TTDMF (combined C-index = 0.84, log rank p < 0.001). Number of smoking pack-years complemented clinical and imaging features only in patients without radiographic extranodal extension or positive retropharyngeal nodes. Time to pattern-specific failure is important for guiding treatment de-escalation strategies, which intend to reduce treatmentrelated toxicity in patients with relatively long expected survival times. This study suggests that PET/CT features should play a crucial role in future de-escalation trials and management of HPV-OPC patients. (C) 2020 Elsevier B.V. All rights reserved.

IMPORTANCE Recent insights into the biologic characteristics and treatment of oropharyngeal cancer may help inform improvements in prognostic modeling. A bayesian multistate model incorporates sophisticated statistical techniques to provide individualized predictions of survival and recurrence outcomes for patients with newly diagnosed oropharyngeal cancer. OBJECTIVE To develop a model for individualized survival, locoregional recurrence, and distant metastasis prognostication for patients with newly diagnosed oropharyngeal cancer, incorporating clinical, oncologic, and imaging data. DESIGN, SETTING, AND PARTICIPANTS In this prognostic study, a data set was used comprising 840 patients with newly diagnosed oropharyngeal cancer treated at a National Cancer Institute-designated center between January 2003 and August 2016; analysis was performed between January 2019 and June 2020. Using these data, a bayesian multistate model was developed that can be used to obtain individualized predictions. The prognostic performance of the model was validated using data from 447 patients treated for oropharyngeal cancer at Erasmus Medical Center in the Netherlands. EXPOSURES Clinical/oncologic factors and imaging biomarkers collected at or before initiation of first-line therapy. MAIN OUTCOMES AND MEASURES Overall survival, locoregional recurrence, and distant metastasis after first-line cancer treatment. RESULTS Of the 840 patients included in the National Cancer Institute-designated center, 715 (85.1%) were men and 268 (31.9%) were current smokers. The Erasmus Medical Center cohort comprised 300 (67.1%) men, with 350 (78.3%) current smokers. Model predictions for 5-year overall survival demonstrated good discrimination, with area under the curve values of 0.81 for the model with and 0.78 for the model without imaging variables. Application of the model without imaging data in the independent Dutch validation cohort resulted in an area under the curve of 0.75. This model possesses good calibration and stratifies patients well in terms of likely outcomes among many competing events. CONCLUSIONS AND RELEVANCE In this prognostic study, a multistate model of oropharyngeal cancer incorporating imaging biomarkers appeared to estimate and discriminate locoregional recurrence from distant metastases. Providing personalized predictions of multiple outcomes increases the information available for patients and clinicians. The web-based application designed in this study may serve as a useful tool for generating predictions and visualizing likely outcomes for a specific patient.

Imaging evaluation of soft tissue masses is essential for diagnosis, preoperative staging, and post-treatment follow-up. Magnetic resonance imaging plays the major role because of its superior resolution that helps in better tissue characterization, and its multiplanar imaging capability in evaluation of soft tissue masses. Additional imaging techniques, such as radiographs, computed tomography, positron-emission tomography-CT, radionuclide scintigraphy and ultrasonography, also play vital roles by providing additional information required in management of soft tissue masses. Knowledge of the usefulness and limitations of these imaging techniques is essential for their judicious selection. This article reviews the current role of various imaging techniques in diagnosis, presurgical planning, and post-treatment follow-up of soft tissue masses.

OBJECTIVE The evaluation, treatment, and prognosis of neonatal brachial plexus palsy (NBPP) continues to have many areas of debate, including the use of ancillary testing. Given the continued improvement in imaging, it is important to revisit its utility. Nerve root avulsions have historically been identified by the presence of pseudomeningoceles or visible ruptures. This "all-or-none" definition of nerve root avulsions has many implications for the understanding and management of NBPP, especially as characterization of the proximal nerve root as a potential donor remains critical. This study examined the ability of high-resolution MRI to more specifically define the anatomy of nerve root avulsions by individually examining the ventral and dorsal rootlets as they exit the spinal cord. METHODS This is a retrospective review of patients who had undergone brachial plexus protocol MRI for clinical evaluation of NBPP at a single institution. Each MR image was independently reviewed by a board-certified neuroradiologist, who was blinded to both established diagnosis/surgical findings and laterality. Each dorsal and ventral nerve rootlet bilaterally from C5 to T1 was evaluated from the spinal cord to its exit in the neuroforamen. Each rootlet was classified as avulsed, intact, or undeterminable. RESULTS Sixty infants underwent brachial plexus protocol MRI from 2010 to 2018. All infants were included in this study. Six hundred individual rootlets were analyzed. There were 49 avulsed nerve rootlets in this cohort. Twenty-nine (59%) combined dorsal/ventral avulsions involved both the ventral and dorsal rootlets, and 20 (41%) were either isolated ventral or isolated dorsal rootlet avulsions. Of the isolated avulsion injuries, 13 (65%) were dorsal only, meaning that the motor rootlets were intact. CONCLUSIONS A closer look at nerve root avulsions with MRI demonstrates a significant prevalence (approximately 41%) of isolated dorsal or ventral nerve rootlet disruptions. This finding implies that nerve roots previously labeled as "avulsed" but with only isolated dorsal (sensory) rootlet avulsion can yet provide donor fascicles in reconstruction strategies. A majority (99%) of the rootlets can be clearly visualized with MRI. These findings may significantly impact the clinical understanding of neonatal brachial plexus injury and its treatment.