Patrick W. Hales

BACKGROUND AND PURPOSE: Pediatric posterior fossa tumors often present with hydrocephalus; postoperatively, up to 25% of patients develop cerebellar mutism syndrome. Arterial spin-labeling is a noninvasive means of quantifying CBF and bolus arrival time. The aim of this study was to investigate how changes in perfusion metrics in children with posterior fossa tumors are modulated by cerebellar mutism syndrome and hydrocephalus requiring pre-resection CSF diversion. MATERIALS AND METHODS: Forty-four patients were prospectively scanned at 3 time points (preoperatively, postoperatively, and at 3-month follow-up) with single- and multi-inflow time arterial spin-labeling sequences. Regional analyses of CBF and bolus arrival time were conducted using coregistered anatomic parcellations. ANOVA and multivariable, linear mixed-effects modeling analysis approaches were used. The study was registered at clinicaltrials.gov (NCT03471026). RESULTS: CBF increased after tumor resection and at follow-up scanning (P = .045). Bolus arrival time decreased after tumor resection and at follow-up scanning (P = .018). Bolus arrival time was prolonged (P = .058) following the midline approach, compared with cerebellar hemispheric surgical approaches to posterior fossa tumors. Multivariable linear mixed-effects modeling showed that regional perfusion changes were more pronounced in the 6 children who presented with symptomatic obstructive hydrocephalus requiring pre-resection CSF diversion, with hydrocephalus lowering the baseline mean CBF by 20.5 (standard error, 6.27) mL/100g/min. Children diagnosed with cerebellar mutism syndrome (8/44, 18.2%) had significantly higher CBF at follow-up imaging than those who were not (P = .040), but no differences in pre- or postoperative perfusion parameters were seen. CONCLUSIONS: Multi-inflow time arterial spin-labeling shows promise as a noninvasive tool to evaluate cerebral perfusion in the setting of pediatric obstructive hydrocephalus and demonstrates increased CBF following resolution of cerebellar mutism syndrome.

Background: High‐risk neuroblastoma (HR‐NB) has a variable response to preoperative chemotherapy. It is not possible to differentiate viable vs. nonviable residual tumor before surgery.Purpose: To explore the association between apparent diffusion coefficient (ADC) values from diffusion‐weighted magnetic resonance imaging (DW‐MRI), ¹²³I‐meta‐iodobenzyl‐guanidine (¹²³I‐mIBG) uptake, and histology before and after chemotherapy.Study Type: Retrospective.Subjects: Forty patients with HR‐NB.Field Strength/Sequence: 1.5T axial DW‐MRI (b = 0,1000 s/mm²) and T₂‐weighted sequences. ¹²³I‐mIBG scintigraphy planar imaging (all patients), with additional ¹²³I‐mIBG single‐photon emission computed tomography / computerized tomography (SPECT/CT) imaging (15 patients).Assessment: ADC maps and ¹²³I‐mIBG SPECT/CT images were coregistered to the T₂‐weighted images. ¹²³I‐mIBG uptake was normalized with a tumor‐to‐liver count ratio (TLCR). Regions of interest (ROIs) for primary tumor volume and different intratumor subregions were drawn. The lower quartile ADC value (ADC_25prc) was used over the entire tumor volume and the overall level of ¹²³I‐mIBG uptake was graded into avidity groups.Statistical Tests: Analysis of variance (ANOVA) and linear regression were used to compare ADC and MIBG values before and after treatment. Threshold values to classify tumors as viable/necrotic were obtained using ROC analysis of ADC and TLCR values.Results: No significant difference in whole‐tumor ADC_25prc values were found between different ¹²³I‐mIBG avidity groups pre‐ (P = 0.31) or postchemotherapy (P = 0.35). In the “intratumor” analysis, 5/15 patients (prechemotherapy) and 0/14 patients (postchemotherapy) showed a significant correlation between ADC and TLCR values (P < 0.05). Increased tumor shrinkage was associated with lower pretreatment tumor ADC_25prc values (P < 0.001); no association was found with pretreatment ¹²³I‐mIBG avidity (P = 0.17). Completely nonviable tumors had significantly lower postchemotherapy ADC_25prc values than tumors with >10% viable tumor (P < 0.05). Both pre‐ and posttreatment TLCR values were significantly higher in patients with >50% viable tumor than those with 10–50% viable tumor (P < 0.05).Data Conclusion: ¹²³I‐mIBG avidity and ADC values are complementary noninvasive biomarkers of therapeutic response in HR‐NB.Level of Evidence: 4.Technical Efficacy Stage: 3.

Objective: ADC (Apparent Diffusion Coefficient) derived from Diffusion-Weighted Imaging (DWI) has shown promise as a non-invasive quantitative imaging biomarker in Wilms’ tumours. However, many non-Gaussian models could be applied to DWI. This study aimed to compare the suitability of four diffusion models (mono exponential, IVIM [Intravoxel Incoherent Motion], stretched exponential, and kurtosis) in Wilms’ tumours and the unaffected contralateral kidneys. Materials and methods: DWI data were retrospectively reviewed (110 Wilms’ tumours and 75 normal kidney datasets). The goodness of fit for each model was measured voxel-wise using Akaike Information Criteria (AIC). Mean AIC was calculated for each tumour volume (or contralateral normal kidney tissue). One-way ANOVAs with Greenhouse–Geisser correction and post hoc tests using the Bonferroni correction evaluated significant differences between AIC values; the lowest AIC indicating the optimum model. Results: IVIM and stretched exponential provided the best fits to the Wilms’ tumour DWI data. IVIM provided the best fit for the normal kidney data. Mono exponential was the least appropriate fitting method for both Wilms’ tumour and normal kidney data. Discussion: The diffusion weighted signal in Wilms’ tumours and normal kidney tissue does not exhibit a mono-exponential decay and is better described by non-Gaussian models of diffusion.

Background: Arterial spin labeling (ASL) is a useful tool for measuring cerebral blood flow (CBF). However, due to the low signal‐to‐noise ratio (SNR) of the technique, multiple repetitions are required, which results in prolonged scan times and increased susceptibility to artifacts.Purpose: To develop a deep‐learning‐based algorithm for simultaneous denoising and suppression of transient artifacts in ASL images.Study Type: Retrospective.Subjects: 131 pediatric neuro‐oncology patients for model training and 11 healthy adult subjects for model evaluation.Field Strength/Sequence: 3T / pseudo‐continuous and pulsed ASL with 3D gradient‐and‐spin‐echo readout.Assessment: A denoising autoencoder (DAE) model was designed with stacked encoding/decoding convolutional layers. Reference standard images were generated by averaging 10 pairwise ASL subtraction images. The model was trained to produce perfusion images of a similar quality using a single subtraction image. Performance was compared against Gaussian and non‐local means (NLM) filters. Evaluation metrics included SNR, peak SNR (PSNR), and structural similarity index (SSIM) of the CBF images, compared to the reference standard.Statistical Tests: One‐way analysis of variance (ANOVA) tests for group comparisons.Results: The DAE model was the only model to produce a significant increase in SNR compared to the raw images (P < 0.05), providing an average SNR gain of 62%. The DAE model was also effective at suppressing transient artifacts, and was the only model to show a significant improvement in accuracy in the generated CBF images, as assessed using PSNR values (P < 0.05). In addition, using data from multiple inflow time acquisitions, the DAE images produced the best fit to the Buxton kinetic model, offering a 75% reduction in the fitting error compared to the raw images.Data Conclusion: Deep‐learning‐based algorithms provide superior accuracy when denoising ASL images, due to their ability to simultaneously increase SNR and suppress artifactual signals in raw ASL images.Level of Evidence: 3Technical Efficacy Stage: 1

Many MRI techniques require prior knowledge of the T1-relaxation time of blood ( T_{1 bl}). An assumed/fixed value is often used; however, T_{1 bl} is sensitive to magnetic field ( B₀), haematocrit ( Hct), and oxygen saturation ( Y). We aimed to combine data from previous in vitro measurements into a mathematical model, to estimate T_{1 bl} as a function of B₀, Hct, and Y. The model was shown to predict T_{1 bl} from in vivo studies with a good accuracy (±87 ms). This model allows for improved estimation of T_{1 bl} between 1.5–7.0 T while accounting for variations in Hct and Y, leading to improved accuracy of MRI-derived perfusion measurements.

Wilms’ tumours (WTs) are large heterogeneous tumours, which typically consist of a mixture of histological cell types, together with regions of chemotherapy‐induced regressive change and necrosis. The predominant cell type in a WT is assessed histologically following nephrectomy, and used to assess the tumour subtype and potential risk. The purpose of this study was to develop a mathematical model to identify subregions within WTs with distinct cellular environments in vivo, determined using apparent diffusion coefficient (ADC) values from diffusion‐weighted imaging (DWI). We recorded the WT subtype from the histopathology of 32 tumours resected in patients who received DWI prior to surgery after pre‐operative chemotherapy had been administered. In 23 of these tumours, DWI data were also available prior to chemotherapy. Histograms of ADC values were analysed using a multi‐Gaussian model fitting procedure, which identified ‘subpopulations’ with distinct cellular environments within the tumour volume. The mean and lower quartile ADC values of the predominant viable tissue subpopulation (ADC_1MEAN, ADC_1LQ), together with the same parameters from the entire tumour volume (ADC_0MEAN, ADC_0LQ), were tested as predictors of WT subtype. ADC_1LQ from the multi‐Gaussian model was the most effective parameter for the stratification of WT subtype, with significantly lower values observed in high‐risk blastemal‐type WTs compared with intermediate‐risk stromal, regressive and mixed‐type WTs (p < 0.05). No significant difference in ADC_1LQ was found between blastemal‐type and intermediate‐risk epithelial‐type WTs. The predominant viable tissue subpopulation in every stromal‐type WT underwent a positive shift in ADC_1MEAN after chemotherapy. Our results suggest that our multi‐Gaussian model is a useful tool for differentiating distinct cellular regions within WTs, which helps to identify the predominant histological cell type in the tumour in vivo. This shows potential for improving the risk‐based stratification of patients at an early stage, and for guiding biopsies to target the most malignant part of the tumour. Copyright © 2015 John Wiley & Sons, Ltd.