Performance of deep learning synthetic CTs for MR-only brain radiation therapy.

PURPOSE: To evaluate the dosimetric and image-guided radiation therapy (IGRT) performance of a novel generative adversarial network (GAN) generated synthetic CT (synCT) in the brain and compare its performance for clinical use including conventional brain radiotherapy, cranial stereotactic radiosurgery (SRS), planar, and volumetric IGRT. METHODS AND MATERIALS: SynCT images for 12 brain cancer patients (6 SRS, 6 conventional) were generated from T1-weighted postgadolinium magnetic resonance (MR) images by applying a GAN model with a residual network (ResNet) generator and a convolutional neural network (CNN) with 5 convolutional layers as the discriminator that classified input images as real or synthetic. Following rigid registration, clinical structures and treatment plans derived from simulation CT (simCT) images were transferred to synCTs. Dose was recalculated for 15 simCT/synCT plan pairs using fixed monitor units. Two-dimensional (2D) gamma analysis (2%/2 mm, 1%/1 mm) was performed to compare dose distributions at isocenter. Dose-volume histogram (DVH) metrics (D95% , D99% , D0.2cc, and D0.035cc ) were assessed for the targets and organ at risks (OARs). IGRT performance was evaluated via volumetric registration between cone beam CT (CBCT) to synCT/simCT and planar registration between KV images to synCT/simCT digital reconstructed radiographs (DRRs). RESULTS: Average gamma passing rates at 1%/1mm and 2%/2mm were 99.0 ± 1.5% and 99.9 ± 0.2%, respectively. Excellent agreement in DVH metrics was observed (mean difference ≤0.10 ± 0.04 Gy for targets, 0.13 ± 0.04 Gy for OARs). The population averaged mean difference in CBCT-synCT registrations were <0.2 mm and 0.1 degree different from simCT-based registrations. The mean difference between kV-synCT DRR and kV-simCT DRR registrations was <0.5 mm with no statistically significant differences observed (P > 0.05). An outlier with a large resection cavity exhibited the worst-case scenario. CONCLUSION: Brain GAN synCTs demonstrated excellent performance for dosimetric and IGRT endpoints, offering potential use in high precision brain cancer therapy.

Impact of CT reconstruction algorithm on auto-segmentation performance.

Model-based iterative reconstruction (MBIR) reduces CT imaging dose while maintaining image quality. However, MBIR reduces noise while preserving edges which may impact intensity-based tasks such as auto-segmentation. This work evaluates the sensitivity of an auto-contouring prostate atlas across multiple MBIR reconstruction protocols and benchmarks the results against filtered back projection (FBP). Images were created from raw projection data for 11 prostate cancer cases using FBP and nine different MBIR reconstructions (3 protocols/3 noise reduction levels) yielding 10 reconstructions/patient. Five bony structures, bladder, rectum, prostate, and seminal vesicles (SVs) were segmented using an auto-segmentation pipeline that renders 3D binary masks for analysis. Performance was evaluated for volume percent difference (VPD) and Dice similarity coefficient (DSC), using FBP as the gold standard. Nonparametric Friedman tests plus post hoc all pairwise comparisons were employed to test for significant differences (P < 0.05) for soft tissue organs and protocol/level combinations. A physician performed qualitative grading of 396 MBIR contours across the prostate, bladder, SVs, and rectum in comparison to FBP using a six-point scale. MBIR contours agreed with FBP for bony anatomy (DSC ≥ 0.98), bladder (DSC ≥ 0.94, VPD < 8.5%), and prostate (DSC = 0.94 ± 0.03, VPD = 4.50 ± 4.77% (range: 0.07-26.39%). Increased variability was observed for rectum (VPD = 7.50 ± 7.56% and DSC = 0.90 ± 0.08) and SVs (VPD and DSC of 8.23 ± 9.86% range (0.00-35.80%) and 0.87 ± 0.11, respectively). Over the all protocol/level comparisons, a significant difference was observed for the prostate VPD between BSPL1 and BSTL2 (adjusted P-value = 0.039). Nevertheless, 300 of 396 (75.8%) of the four soft tissue structures using MBIR were graded as equivalent or better than FBP, suggesting that MBIR offered potential improvements in auto-segmentation performance when compared to FBP. Future work may involve tuning organ-specific MBIR parameters to further improve auto-segmentation performance. Running title: Impact of CT Reconstruction Algorithm on Auto-segmentation Performance.

Geometric and dosimetric impact of anatomical changes for MR-only radiation therapy for the prostate.

PURPOSE: With the move towards magnetic resonance imaging (MRI) as a primary treatment planning modality option for men with prostate cancer, it becomes critical to quantify the potential uncertainties introduced for MR-only planning. This work characterized geometric and dosimetric intra-fractional changes between the prostate, seminal vesicles (SVs), and organs at risk (OARs) in response to bladder filling conditions. MATERIALS AND METHODS: T2-weighted and mDixon sequences (3-4 time points/subject, at 1, 1.5 and 3.0 T with totally 34 evaluable time points) were acquired in nine subjects using a fixed bladder filling protocol (bladder void, 20 oz water consumed pre-imaging, 10 oz mid-session). Using mDixon images, Magnetic Resonance for Calculating Attenuation (MR-CAT) synthetic computed tomography (CT) images were generated by classifying voxels as muscle, adipose, spongy, and compact bone and by assignment of bulk Hounsfield Unit values. Organs including the prostate, SVs, bladder, and rectum were delineated on the T2 images at each time point by one physician. The displacement of the prostate and SVs was assessed based on the shift of the center of mass of the delineated organs from the reference state (fullest bladder). Changes in dose plans at different bladder states were assessed based on volumetric modulated arc radiotherapy (VMAT) plans generated for the reference state. RESULTS: Bladder volume reduction of 70 ± 14% from the final to initial time point (relative to the final volume) was observed in the subject population. In the empty bladder condition, the dose delivered to 95% of the planning target volume (PTV) (D95%) reduced significantly for all cases (11.53 ± 6.00%) likely due to anterior shifts of prostate/SVs relative to full bladder conditions. D15% to the bladder increased consistently in all subjects (42.27 ± 40.52%). Changes in D15% to the rectum were patient-specific, ranging from -23.93% to 22.28% (-0.76 ± 15.30%). CONCLUSIONS: Variations in the bladder and rectal volume can significantly dislocate the prostate and OARs, which can negatively impact the dose delivered to these organs. This warrants proper preparation of patients during treatment and imaging sessions, especially when imaging required longer scan times such as MR protocols.

An extended vascular model for less biased estimation of permeability parameters in DCE-T1 images.

One of the key elements in dynamic contrast enhanced (DCE) image analysis is the arterial input function (AIF). Traditionally, in DCE studies a global AIF sampled from a major artery or vein is used to estimate the vascular permeability parameters; however, not addressing dispersion and delay of the AIF at the tissue level can lead to biased estimates of these parameters. To find less biased estimates of vascular permeability parameters, a vascular model of the cerebral vascular system is proposed that considers effects of dispersion of the AIF in the vessel branches, as well as extravasation of the contrast agent (CA) to the extravascular-extracellular space. Profiles of the CA concentration were simulated for different branching levels of the vascular structure, combined with the effects of vascular leakage. To estimate the permeability parameters, the extended model was applied to these simulated signals and also to DCE-T1 (dynamic contrast enhanced T1 ) images of patients with glioblastoma multiforme tumors. The simulation study showed that, compared with the case of solving the pharmacokinetic equation with a global AIF, using the local AIF that is corrected by the vascular model can give less biased estimates of the permeability parameters (Ktrans , vp and Kb ). Applying the extended model to signals sampled from different areas of the DCE-T1 image showed that it is able to explain the CA concentration profile in both the normal areas and the tumor area, where effects of vascular leakage exist. Differences in the values of the permeability parameters estimated in these images using the local and global AIFs followed the same trend as the simulation study. These results demonstrate that the vascular model can be a useful tool for obtaining more accurate estimation of parameters in DCE studies.

A parametric model of the brain vascular system for estimation of the arterial input function (AIF) at the tissue level.

In this paper, we introduce a novel model of the brain vascular system, which is developed based on laws of fluid dynamics and vascular morphology. This model is used to address dispersion and delay of the arterial input function (AIF) at different levels of the vascular structure and to estimate the local AIF in DCE images. We developed a method based on the simplex algorithm and Akaike information criterion to estimate the likelihood of the contrast agent concentration signal sampled in DCE images belonging to different layers of the vascular tree or being a combination of different signal levels from different nodes of this structure. To evaluate this method, we tested the method on simulated local AIF signals at different levels of this structure. Even down to a signal to noise ratio of 5.5 our method was able to accurately detect the branching level of the simulated signals. When two signals with the same power level were combined, our method was able to separate the base signals of the composite AIF at the 50% threshold. We applied this method to dynamic contrast enhanced computed tomography (DCE-CT) data, and using the parameters estimated by our method we created an arrival time map of the brain. Our model corrected AIF can be used for solving the pharmacokinetic equations for more accurate estimation of vascular permeability parameters in DCE imaging studies.

Optimization of a novel large field of view distortion phantom for MR-only treatment planning.

PURPOSE: MR-only treatment planning requires images of high geometric fidelity, particularly for large fields of view (FOV). However, the availability of large FOV distortion phantoms with analysis software is currently limited. This work sought to optimize a modular distortion phantom to accommodate multiple bore configurations and implement distortion characterization in a widely implementable solution. METHOD AND MATERIALS: To determine candidate materials, 1.0 T MR and CT images were acquired of twelve urethane foam samples of various densities and strengths. Samples were precision-machined to accommodate 6 mm diameter paintballs used as landmarks. Final material candidates were selected by balancing strength, machinability, weight, and cost. Bore sizes and minimum aperture width resulting from couch position were tabulated from the literature (14 systems, 5 vendors). Bore geometry and couch position were simulated using MATLAB to generate machine-specific models to optimize the phantom build. Previously developed software for distortion characterization was modified for several magnet geometries (1.0 T, 1.5 T, 3.0 T), compared against previously published 1.0 T results, and integrated into the 3D Slicer application platform. RESULTS: All foam samples provided sufficient MR image contrast with paintball landmarks. Urethane foam (compressive strength ∼1000 psi, density ~20 lb/ft3 ) was selected for its accurate machinability and weight characteristics. For smaller bores, a phantom version with the following parameters was used: 15 foam plates, 55 × 55 × 37.5 cm3 (L×W×H), 5,082 landmarks, and weight ~30 kg. To accommodate > 70 cm wide bores, an extended build used 20 plates spanning 55 × 55 × 50 cm3 with 7,497 landmarks and weight ~44 kg. Distortion characterization software was implemented as an external module into 3D Slicer's plugin framework and results agreed with the literature. CONCLUSION: The design and implementation of a modular, extendable distortion phantom was optimized for several bore configurations. The phantom and analysis software will be available for multi-institutional collaborations and cross-validation trials to support MR-only planning.

Measurement of rat brain tumor kinetics using an intravascular MR contrast agent and DCE-MRI nested model selection.

PURPOSE: Using dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in a rat glioma model, and nested model selection (NMS), to compare estimates of the pharmacokinetic parameters vp , K(trans) , and ve for two different contrast agents (CAs)-gadofosveset, which reversibly binds to human serum albumin, and gadopentetate dimeglumine, which does not. MATERIALS AND METHODS: DCE-MRI studies were performed on nine Fisher 344 rats inoculated intracerebrally with 9L gliosarcoma cells using both gadofosveset and gadopentetate. The parameters vp , K(trans) , and ve were estimated using NMS. RESULTS: K(trans) estimates using gadofosveset, compared to gadopentetate, differed in their means (gadofosveset 0.025 ± 0.008 min(-1) vs. gadopentetate 0.046 ± 0.011 min(-1) ; P = 0.0039). This difference notwithstanding, the intraclass correlation coefficient (ICC) for the two estimates of K(trans) showed nearly perfect linear dependence (ICC = 0.8479 by Pearson's r). Other estimates, ve (gadofosveset 22.7 ± 4.7% vs. gadopentetate 23.6 ± 5.6%; P = 0.4258) and vp (gadofosveset 1.5 ± 0.5% vs. gadopentetate 1.6 ± 0.4%; P = 0.25), were not different in their means between the two CAs, and there was almost perfect agreement for ve (ICC = 0.8798) and substantial agreement for vp (ICC = 0.7981) between the two CAs. CONCLUSION: Estimates of K(trans) were statistically different using gadofosveset and gadopentetate, whereas ve and vp were similar with two CAs. NMS produced robust estimates of pharmacokinetic parameters using DCE-MRI that show promise as important measures of tumor physiology and microenvironment.

Validation of daily 0.35 T diffusion-weighted MRI for MRI-guided glioblastoma radiotherapy.

BACKGROUND: MRI-Linac systems enable daily diffusion-weighed imaging (DWI) MRI scans for assessing glioblastoma tumor changes with radiotherapy treatment. PURPOSE: Our study assessed the image quality of echoplanar imaging (EPI)-DWI scans compared with turbo spin echo (TSE)-DWI scans at 0.35 Tesla (T) and compared the apparent diffusion coefficient (ADC) values and distortion of EPI-DWI on 0.35 T MRI-Linac compared to high-field diagnostic MRI scanners. METHODS: The calibrated National Institute of Standards and Technology (NIST)/Quantitative Imaging Biomarkers Alliance (QIBA) Diffusion Phantom was scanned on a 0.35 T MRI-Linac, and 1.5 T and 3 T MRI with EPI-DWI. Five patients were scanned on a 0.35 T MRI-Linac with a TSE-DWI sequence, and five other patients were scanned with EPI-DWI on a 0.35 T MRI-Linac and a 3 T MRI. The quality of images was compared between the TSE-DWI and EPI-DWI on the 0.35 T MRI-Linac assessing signal-to-noise ratios and presence of artifacts. EPI-DWI ADC values and distortion magnitude were measured and compared between 0.35 T MRI-Linac and high-field MRI for both phantom and patient studies. RESULTS: The average ADC differences between EPI-DWI acquired on the 0.35 T MRI-Linac, 1.5 T and 3 T MRI scanners and published references in the phantom study were 1.7%, 0.4% and 1.0%, respectively. Comparing the ADC values based on EPI-DWI in glioblastoma tumors, there was a 3.36% difference between 0.35 and 3 T measurements. Susceptibility-induced distortions in the EPI-DWI phantoms were 0.46 ± 1.51 mm for 0.35 MRI-Linac, 0.98 ± 0.51 mm for 1.5 T MRI and 1.14 ± 1.88 mm for 3 T MRI; for patients -0.47 ± 0.78 mm for 0.35 T and 1.73 ± 2.11 mm for 3 T MRIs. The mean deformable registration distortion for a phantom was 1.1 ± 0.22 mm, 3.5 ± 0.39 mm and 4.7 ± 0.37 mm for the 0.35 T MRI-Linac, 1.5 T MRI, and 3 T MRI scanners, respectively; for patients this distortion was -0.46 ± 0.57 mm for 0.35 T and 4.2 ± 0.41 mm for 3 T. EPI-DWI 0.35 T MRI-Linac images showed higher SNR and lack of artifacts compared with TSE-DWI, especially at higher b-values up to 1000 s/mm2. CONCLUSION: EPI-DWI on a 0.35 T MRI-Linac showed superior image quality compared with TSE-DWI, minor and less distortions than high-field diagnostic scanners, and comparable ADC values in phantoms and glioblastoma tumors. EPI-DWI should be investigated on the 0.35 T MRI-Linac for prediction of early response in patients with glioblastoma.

ASSET: Auto-Segmentation of the Seventeen SEgments for Ventricular Tachycardia Ablation in Radiation Therapy.

There has been a recent effort to treat high-risk ventricular tachycardia (VT) patients through radio-ablation. However, manual segmentation of the VT target is complex and time-consuming. This work introduces ASSET, or Auto-segmentation of the Seventeen SEgments for Tachycardia ablation, to aid in radiation therapy (RT) planning. ASSET was retrospectively applied to CTs for 26 thoracic RT patients (13 undergoing VT ablation). The physician-defined parasternal long-axis of the left ventricle (LV) and the axes generated from principal component analysis (PCA) were compared using mean distance to agreement (MDA) and angle of separation. The manually selected right ventricle insertion point and LVs were used to apply the ASSET model to automatically generate the 17 segments of the LV myocardium (LVM). Physician-defined parasternal long-axis differed from PCA by 1.2 ± 0.3 mm MDA and 6.9 ± 0.7 degrees. Segments differed by 0.69 ± 0.29 mm MDA and 0.89 ± 0.03 Dice similarity coefficient. Running ASSET takes <5 min where manual segmentation took >2 h/patient. Agreement between ASSET and expert contours was comparable to inter-observer variability. Qualitative scoring conducted by three experts revealed automatically generated segmentations were clinically useable as-is. ASSET offers efficient and reliable automatic segmentations for the 17 segments of the LVM for target generation in RT planning.

Model selection for DCE-T1 studies in glioblastoma.

Dynamic contrast enhanced T(1)-weighted MRI using the contrast agent gadopentetate dimeglumine (Gd-DTPA) was performed on 10 patients with glioblastoma. Nested models with as many as three parameters were used to estimate plasma volume or plasma volume and forward vascular transfer constant (K(trans)) and the reverse vascular transfer constant (k(ep)). These constituted models 1, 2, and 3, respectively. Model 1 predominated in normal nonleaky brain tissue, showing little or no leakage of contrast agent. Model 3 predominated in regions associated with aggressive portions of the tumor, and model 2 bordered model 3 regions, showing leakage at reduced rates. In the patient sample, v(p) was about four times that of white matter in the enhancing part of the tumor. K(trans) varied by a factor of 10 between the model 2 (1.9 ↔ 10(-3) min(-1)) and model 3 regions (1.9 ↔ 10(-2) min(-1)). The mean calculated interstitial space (model 3) was 5.5%. In model 3 regions, excellent curve fits were obtained to summarize concentration-time data (mean R(2) = 0.99). We conclude that the three parameters of the standard model are sufficient to fit dynamic contrast enhanced T(1) data in glioblastoma under the conditions of the experiment.

Impairments of white matter tracts and connectivity alterations in five cognitive networks of patients with multiple sclerosis.

INTRODUCTION: MS is associated with structural and functional brain alterations leading to cognitive impairments across multiple domains including attention, memory, and speed of information processing. Here, we analyzed the white matter damage and topological organization of white matter tracts in specific brain regions responsible for cognition in MS. METHODS: Brain DTI, rs-fMRI, T1, T2, and T2-FLAIR were acquired for 22 MS subjects and 22 healthy controls. Automatic brain parcellation was performed on T1-weighted images. Skull-stripped T1-weighted intensity inverted images were co-registered to the b0 image. Diffusion-weighted images were processed to perform whole brain tractography. The rs-fMRI data were processed, and the connectivity matrixes were analyzed to identify significant differences in the network of nodes between the two groups using NBS analysis. In addition, diffusion entropy maps were produced from DTI data sets using in-house software. RESULTS: MS subjects exhibited significantly reduced mean FA and entropy in 38 and 34 regions, respectively, out of a total of 54 regions. The connectivity values in both structural and functional analyses were decreased in most regions of the default mode network and in four other cognitive networks in MS subjects compared to healthy controls. MS also induced significant reduction in the normalized hippocampus and corpus callosum volumes; the normalized hippocampus volume was significantly correlated with EDSS scores. CONCLUSION: MS subjects have significant white matter damage and reduction of FA and entropy in various brain regions involved in cognitive networks. Structural and functional connectivity within the default mode network and an additional four cognitive networks exhibited significant changes compared with healthy controls.

Rapid multicontrast brain imaging on a 0.35T MR-linac.

PURPOSE: Magnetic resonance-guided radiation therapy (MRgRT) has shown great promise for localization and real-time tumor monitoring. However, to date, quantitative imaging has been limited for low field MRgRT. This work benchmarks quantitative T1, R2*, and Proton Density (PD)mapping in a phantom on a 0.35 T MR-linac and implements a novel acquisition method, STrategically Acquired Gradient Echo (STAGE). To further validate STAGE in a clinical setting, a pilot study was undertaken in a cohort of brain tumor patients to elucidate opportunities for longitudinal functional imaging with an MR-linac in the brain. METHODS: STAGE (two triple-echo gradient echo (GRE) acquisitions) was optimized for a 0.35T low-field MR-linac. Simulations were performed to choose two flip angles to optimize signal-to-noise ratio (SNR) and T1-mapping precision. Tradeoffs between SNR, scan time, and spatial resolution for whole-brain coverage were evaluated in healthy volunteers. Data were inputted into a STAGE processing pipeline to yield four qualitative images (T1-weighted, enhanced T1-weighted, proton-density (PD) weighted, and simulated FLuid-Attenuated Inversion Recovery (sFLAIR)), and three quantitative datasets (T1, PD, and R2*). A benchmarking ISMRM/NIST phantom consisting of vials with variable NiCl2 and MnCl2 concentrations was scanned using variable flip angles (VFA) (2-60 degrees) and inversion recovery (IR) methods at 0.35 T. STAGE and VFA T1 values of vials were compared to IR T1 values. As measures of agreement with reference values and repeatability, relative error (RE) and coefficient of variability (CV) were calculated, respectively, for quantitative MR values within the phantom vials (spheres). To demonstrate feasibility, longitudinal STAGE data (pretreatment, weekly, and ~ 2 months post-treatment) were acquired in an IRB-approved pilot study of brain tumor cases via the generation of temporal and differential quantitative MRI maps. RESULTS: In the phantom, RE of measured VFA T1 and STAGE relative to IR reference values were 7.0 ± 2.5% and 9.5 ± 2.2% respectively. RE for the PD vials was 8.1 ± 6.8% and CV for phantom R2* measurements was 10.1 ± 9.9%. Simulations and volunteer experiments yielded final STAGE parameters of FA = 50°/10°, 1 × 1 × 3 mm3 resolution, TR = 40 ms, TE = 5/20/34 ms in 10 min (64 slices). In the pilot study of brain tumor patients, differential maps for R2* and T1 maps were sensitive to local tumor changes and appeared similar to 3 T follow-up MRI datasets. CONCLUSION: Quantitative T1, R2*, and PD mapping are promising at 0.35 T agreeing well with reference data. STAGE phantom data offer quantitative representations comparable to traditional methods in a fraction of the acquisition time. Initial feasibility of implementing STAGE at 0.35 T in a patient brain tumor cohort suggests that detectable changes can be observed over time. With confirmation in a larger cohort, results may be implemented to identify areas of recurrence and facilitate adaptive radiation therapy.

Large field of view distortion assessment in a low-field MR-linac.

PURPOSE: MR-guided radiation therapy (RT) offers unparalleled soft tissue contrast for localization and target tracking. However, MRI distortions may be detrimental to high precision RT. This work characterizes the gradient nonlinearity (GNL) and total distortions over the first year of clinical operation of a 0.35T MR-linac. METHODS: For GNL characterization, an in-house large field of view (FOV) phantom (60 × 42.5 × 55 cm3 , >6000 spherical landmarks) was configured and scanned at four timepoints with forward/reverse read polarities (Gradient Echo sequence, FA/TR/TE = 28°/30 ms/6 ms). GNL was measured in Anterior-Posterior (AP), Left-Right (LR), and Superior-Inferior (SI) frequency-encoding directions based on deviation of the auto-segmented landmark centroids between rigidly registered MR and CT images and assessed based on radial distance from magnet isocenter. Total distortion was assessed using a 30 × 30 cm2 grid phantom oriented along the cardinal axes over >1 year of operation. RESULTS: The scanner's spatial integrity within the first ~10 months was stable (maximum total distortion variation = 10/6/8%, maximum distortion = 1.41/0.99/1.56 mm in Axial/Coronal/Sagittal planes, respectively). GNL distortions measured during this time period <10 cm from isocenter were (-0.74, 0.45), (-0.67, 0.53), and (-0.86, 0.70) mm in AP/LR/SI directions. In the 10-20 cm range, <1.5% of the distortions exceeded 2 mm in the AP and LR axes while <4% of the distortions exceeded 2 mm for SI. After major repairs and magnet re-shim, detectable changes were observed in total and GNL distortions (20% reduction in AP and 36% increase in SI direction in the 20-25 cm range). Across all timepoints and axes, 38-53% of landmarks in the 20-25 cm range were displaced by >1 mm. CONCLUSIONS: GNL distortions were negligible within a 10 cm radius from isocenter. However, in the periphery, non-negligible distortions of up to ~7 mm were observed, which may necessitate GNL corrections for MR-IGRT for treatment sites distant from magnet isocenter.

A novel and rapid approach to estimate patient-specific distortions based on mDIXON MRI.

While MRI-only radiation treatment planning (RTP) is becoming more widespread, a robust clinical solution for patient-specific distortion corrections is not available. This work explores B 0 mapping based on mDIXON imaging, often performed for MR-only RTP, as an alternative to separate dual-acquisition gradient-recalled echo imaging, with the overarching goal of developing an efficient and robust approach for patient-specific distortion correction. Initial benchmarking was conducted by scanning a phantom and generating B 0 field maps with two approaches: (1) conventional B 0 mapping and (2) experimental mDIXON imaging. Distortion maps were derived from the field maps and compared. The head and neck regions, including brain, of ten healthy volunteers were then evaluated at 1.5 T and 3 T. Distortion maps were again compared between approaches, using difference maps and histogram analysis. Overall, conventional B 0 mapping was well approximated by mDIXON imaging: The distortions of 95% of the voxels in the phantom estimated by mDIXON and conventional B 0 mapping differed by <0.02 mm (1.5 T) and <0.04 mm (3 T), while the 95-percentiles of the distortions estimated by conventional B 0 mapping were <0.06 mm (1.5 T) and <0.12 mm (3 T). In head and neck the distortions of 99% of the voxels were within ±0.2 mm at 1.5 T for both approaches and within ±0.4 mm and ±0.5 mm at 3 T for mDIXON imaging and conventional B 0 mapping, respectively. The majority of differences in vivo were confined to regions with high spatial variation of the B 0 field, mostly around internal air cavities. For 1.5 T, the mDIXON imaging-based correction alone reduced the 95-percentile of distortions from 0.15 mm to 0.03 mm and within the brain from 0.06 mm to 0.02 mm. Slightly lower reductions were observed at 3 T. In conclusion, mDIXON imaging closely approximated conventional B 0 mapping for patient-specific distortion assessment. Estimates in the brain were in good agreement, and slight differences were observed near air/tissue interfaces in the head and neck. Overall, mDIXON imaging-based B 0 field maps may be advantageous for rapid patient-specific distortion correction without additional imaging.

Generating synthetic CTs from magnetic resonance images using generative adversarial networks.

PURPOSE: While MR-only treatment planning using synthetic CTs (synCTs) offers potential for streamlining clinical workflow, a need exists for an efficient and automated synCT generation in the brain to facilitate near real-time MR-only planning. This work describes a novel method for generating brain synCTs based on generative adversarial networks (GANs), a deep learning model that trains two competing networks simultaneously, and compares it to a deep convolutional neural network (CNN). METHODS: Post-Gadolinium T1-Weighted and CT-SIM images from fifteen brain cancer patients were retrospectively analyzed. The GAN model was developed to generate synCTs using T1-weighted MRI images as the input using a residual network (ResNet) as the generator. The discriminator is a CNN with five convolutional layers that classified the input image as real or synthetic. Fivefold cross-validation was performed to validate our model. GAN performance was compared to CNN based on mean absolute error (MAE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR) metrics between the synCT and CT images. RESULTS: GAN training took ~11 h with a new case testing time of 5.7 ± 0.6 s. For GAN, MAEs between synCT and CT-SIM were 89.3 ± 10.3 Hounsfield units (HU) and 41.9 ± 8.6 HU across the entire FOV and tissues, respectively. However, MAE in the bone and air was, on average, ~240-255 HU. By comparison, the CNN model had an average full FOV MAE of 102.4 ± 11.1 HU. For GAN, the mean PSNR was 26.6 ± 1.2 and SSIM was 0.83 ± 0.03. GAN synCTs preserved details better than CNN, and regions of abnormal anatomy were well represented on GAN synCTs. CONCLUSIONS: We developed and validated a GAN model using a single T1-weighted MR image as the input that generates robust, high quality synCTs in seconds. Our method offers strong potential for supporting near real-time MR-only treatment planning in the brain.

Per-organ assessment of subject-induced susceptibility distortion for MR-only male pelvis treatment planning.

BACKGROUND: Patient-specific distortions, particularly near tissue/air interfaces, require assessment for magnetic resonance (MR) only radiation treatment planning (RTP). However, patients are dynamic due to changes in physiological status during imaging sessions. This work investigated changes in subject-induced susceptibility distortions to pelvic organs at different bladder states to support pelvis MR-only RTP. METHODS: Pelvises of 9 healthy male volunteers were imaged at 1.0 Tesla (T), 1.5 T, and 3.0 T. Subject-induced susceptibility distortion field maps were generated using a dual-echo gradient-recalled echo (GRE) sequence with B0 field maps obtained from the phase difference between the two echoes acquired at several bladder volume states (3-4/subject, 32 overall). T2 turbo spin echo images were also acquired at each bladder state for organ delineation. Magnet central frequency was tracked over time. Distortion map differences and boxplots were computed to characterize changes within the clinical target volume (CTV), bladder, seminal vesicles, and prostate volumes. RESULTS: The time between the initial and final B0 maps was 42.6 ± 13.9 (range: 13.2-62.1) minutes with minimal change in magnet central frequency (0.02 ± 0.05 mm (range: - 0.06 - 0.12 mm)). Subject-induced susceptibility distortion across all bladder states, field strengths, and subjects was relatively small (1.4-1.9% of all voxels in the prostate and seminal vesicles were distorted > 0.5 mm). In the bladder, no voxels exhibited distortions > 1 mm. An extreme case acquired at 3.0 T with a large volume of rectal air yielded 27.4-34.6% of voxels within the CTVs had susceptibility-induced distortions > 0.5 mm across all time points. CONCLUSIONS: Our work suggests that subject-induced susceptibility distortions caused by bladder/rectal conditions are generally small and subject-dependent. Local changes may be non-negligible within the CTV, thus proper management of filling status is warranted. Future work evaluating the impact of multiple models to accommodate for extreme status changes may be advantageous.

Diffusion-Derived Magnetic Resonance Imaging Measures of Longitudinal Microstructural Remodeling Induced by Marrow Stromal Cell Therapy after Traumatic Brain Injury.

Using magnetic resonance imaging (MRI) and an animal model of traumatic brain injury (TBI), we investigated the capacity and sensitivity of diffusion-derived measures, fractional anisotropy (FA), and diffusion entropy, to longitudinally identify structural plasticity in the injured brain in response to the transplantation of human bone marrow stromal cells (hMSCs). Male Wistar rats (300-350g, n = 30) were subjected to controlled cortical impact TBI. At 6 h or 1 week post-injury, these rats were intravenously injected with 1 mL of saline (at 6 h or 1 week, n = 5/group) or with hMSCs in suspension (∼3 × 106 hMSCs, at 6 h or 1 week, n = 10/group). In vivo MRI measurements and sensorimotor function estimates were performed on all animals pre-injury, 1 day post-injury, and weekly for 3 weeks post-injury. Bielschowsky's silver and Luxol fast blue staining were used to reveal the axon and myelin status, respectively, with and without cell treatment after TBI. Based on image data and histological observation, regions of interest encompassing the structural alterations were made and the values of FA and entropy were monitored in these specific brain regions. Our data demonstrate that administration of hMSCs after TBI leads to enhanced white matter reorganization particularly along the boundary of contusional lesion, which can be identified by both FA and entropy. Compared with the therapy performed at 1 week post-TBI, cell intervention executed at 6 h expedites the brain remodeling process and results in an earlier functional recovery. Although FA and entropy present a similar capacity to dynamically detect the microstructural changes in the tissue regions with predominant orientation of fiber tracts, entropy exhibits a sensitivity superior to that of FA, in probing the structural alterations in the tissue areas with complex fiber patterns.

Resting state fMRI connectivity analysis as a tool for detection of abnormalities in five different cognitive networks of the brain in Multiple Sclerosis patients.

OBJECTIVES: Cognitive dysfunction is present in at least half of patients with Multiple Sclerosis. The purpose of this study was to examine functional connectivity abnormalities in patients with multiple sclerosis (MS) using resting state fMRI (rsfMRI). METHODS: Conventional MRI, rsfMRI and diffusion tensor imaging (DTI) data was acquired from 10 patients with relapsing-remitting multiple sclerosis (RRMS) and 20 healthy controls. Cross-correlation of the resting state average signal among the voxels in each brain region of the five cognitive networks: default mode network (DMN), attention, verbal memory, memory, and visuospatial working memory network, was calculated. Voxelwise analyses were used to investigate fractional anisotropy (FA) of white matter tracts. The normalized gray matter (GM), white matter and thalamus volumes were calculated. RESULTS: Compared to controls, significant deficit in MS patients at each of five networks, attention (p=0.026), DMN (p=0.004), verbal memory (p<0.001), memory (p=0.001), visuospatial working memory (p=0.003) was found. Significant reduction (p=0.034) in the normalized GM volume and asymmetry in thalamus volume (p=0.041) was detected in MS patients compared to controls. CONCLUSION: Wide spread of functional abnormalities are present within different cognitive networks in patients with RRMS, suggesting that DMN may not be sufficient for measurement of MS cognitive impairment. Larger and longitudinal studies should ascertain whether rsfMRI of cognitive networks and changes in GM and thalamus volume can be used as tools for assessment of cognition in clinical trials in MS.

Stroke increases neural stem cells and angiogenesis in the neurogenic niche of the adult mouse.

The unique cellular and vascular architecture of the adult ventricular-subventricular zone (V/SVZ) neurogenic niche plays an important role in regulating neural stem cell function. However, the in vivo identification of neural stem cells and their relationship to blood vessels within this niche in response to stroke remain largely unknown. Using whole-mount preparation of the lateral ventricle wall, we examined the architecture of neural stem cells and blood vessels in the V/SVZ of adult mouse over the course of 3 months after onset of focal cerebral ischemia. Stroke substantially increased the number of glial fibrillary acidic protein (GFAP) positive neural stem cells that are in contact with the cerebrospinal fluid (CSF) via their apical processes at the center of pinwheel structures formed by ependymal cells residing in the lateral ventricle. Long basal processes of these cells extended to blood vessels beneath the ependymal layer. Moreover, stroke increased V/SVZ endothelial cell proliferation from 2% in non-ischemic mice to 12 and 15% at 7 and 14 days after stroke, respectively. Vascular volume in the V/SVZ was augmented from 3% of the total volume prior to stroke to 6% at 90 days after stroke. Stroke-increased angiogenesis was closely associated with neuroblasts that expanded to nearly encompass the entire lateral ventricular wall in the V/SVZ. These data indicate that stroke induces long-term alterations of the neural stem cell and vascular architecture of the adult V/SVZ neurogenic niche. These post-stroke structural changes may provide insight into neural stem cell mediation of stroke-induced neurogenesis through the interaction of neural stem cells with proteins in the CSF and their sub-ependymal neurovascular interaction.