Improving the lesion appearance on FLAIR images synthetized from quantitative MRI: a fast, hybrid approach.

OBJECTIVE: The image quality of synthetized FLAIR (fluid attenuated inversion recovery) images is generally inferior to its conventional counterpart, especially regarding the lesion contrast mismatch. This work aimed to improve the lesion appearance through a hybrid methodology. MATERIALS AND METHODS: We combined a full brain 5-min MR-STAT acquisition followed by FLAIR synthetization step with an ultra-under sampled conventional FLAIR sequence and performed the retrospective and prospective analysis of the proposed method on the patient datasets and a healthy volunteer. RESULTS: All performance metrics of the proposed hybrid FLAIR images on patient datasets were significantly higher than those of the physics-based FLAIR images (p < 0.005), and comparable to those of conventional FLAIR images. The small difference between prospective and retrospective analysis on a healthy volunteer demonstrated the validity of the retrospective analysis of the hybrid method as presented for the patient datasets. DISCUSSION: The proposed hybrid FLAIR achieved an improved lesion appearance in the clinical cases with neurological diseases compared to the physics-based FLAIR images, Future prospective work on patient data will address the validation of the method from a diagnostic perspective by radiological inspection of the new images over a larger patient cohort.

Fast and silent MRI using nonlinear gradient fields at the ultrasonic gradient switching frequency of 20 kHz with a Point Spread Function framework reconstruction.

PURPOSE: To demonstrate the feasibility of using a nonlinear gradient field for spatial encoding at the ultrasonic switching frequency of 20 kHz and present a framework to reconstruct data acquired in this way. METHODS: Nonlinear encoding at 20 kHz was realized by using a single-axis silent gradient insert for imaging in the periphery, that, is the nonlinear region, of the gradient field. The gradient insert induces a rapidly oscillating gradient field in the phase-encode direction, which enables nonlinear encoding when combined with a Cartesian readout from the linear whole-body gradients. Data from a 2D gradient echo sequence were reconstructed using a point spread function (PSF) framework. Accelerated scans were also simulated via retrospective undersampling (R = 1 to R = 8) to determine the effectiveness of the PSF-framework for accelerated imaging. RESULTS: Using a nonlinear gradient field switched at 20 kHz and the PSF-framework resulted in images of comparable quality to images from conventional Cartesian linear encoding. At increased acceleration factors (R ≤ 8), the PSF-framework outperformed linear SENSE reconstructions by improved controlling of aliasing artifacts. CONCLUSION: Using the PSF-framework, images of comparable quality to conventional SENSE reconstructions are possible via combining traditional linear and ultrasonic oscillating nonlinear encoding fields. Using nonlinear gradient fields relaxes the demand for strictly linear gradient fields, enabling much higher slew rates with a reduced risk of peripheral nerve stimulation or cardiac stimulation, which could aid in extension to ultrasonic whole-body MRI. The lack of aliasing artifacts also highlights the potential of accelerated imaging using the PSF-framework.

Unrolled Optimization via Physics-Assisted Convolutional Neural Network for MR-Based Electrical Properties Tomography: A Numerical Investigation.

Magnetic Resonance imaging based Electrical Properties Tomography (MR-EPT) is a non-invasive technique that measures the electrical properties (EPs) of biological tissues. In this work, we present and numerically investigate the performance of an unrolled, physics-assisted method for 2D MR-EPT reconstructions, where a cascade of Convolutional Neural Networks is used to compute the contrast update. Each network takes in input the EPs and the gradient descent direction (encoding the physics underlying the adopted scattering model) and returns as output the updated contrast function. The network is trained and tested in silico using 2D slices of realistic brain models at 128 MHz. Results show the capability of the proposed procedure to reconstruct EPs maps with quality comparable to that of the popular Contrast Source Inversion-EPT, while significantly reducing the computational time.

An improved method to measure transfer functions using MRI.

PURPOSE: A previously published method for MRI-based transfer function assessment makes use of the so-called transceive phase assumption (TPA). This limits its applicability to shorter leads and/or lower field strengths. A new method is presented where the background electric field is determined from both B 1 + $$ {\mathrm{B}}_1^{+} $$ - and B 1 - $$ {\mathrm{B}}_1^{-} $$ -field distributions, avoiding the TPA and making it more generally applicable. THEORY AND METHODS: These B 1 $$ {\mathrm{B}}_1 $$ -distributions are determined from a spoiled gradient echo multiflip angle acquisition. From the separated B 1 $$ {\mathrm{B}}_1 $$ -components the background electrical field and the induced current are computed. Further improvement is achieved by recasting the B 1 $$ {\mathrm{B}}_1 $$ -field model as a "magnitude squared least squares" problem. The proposed reconstruction method is used to determine transfer functions of various copper wire lengths up to 40 cm inside an elliptical ASTM phantom. The method is first tested on EM-simulated data and subsequently phantom and bench measurements are used to determine transfer functions experimentally. RESULTS: In silica reconstructions demonstrate the validity of the proposed B 1 $$ {\mathrm{B}}_1 $$ -field model resulting in highly accurate reconstructed B 1 $$ {\mathrm{B}}_1 $$ -fields, currents, incident electric fields and transfer functions. The experimental results show slight deviations in the field model, however, resulting transfer functions are accurately determined with high similarity to simulations and comparable to bench measurements. CONCLUSION: A more generally applicable method for MRI-based transfer function assessment is presented. The proposed method circumvents phase assumptions making it applicable for longer objects and/or higher field strengths. Additional improvements are implemented in the B 1 $$ {\mathrm{B}}_1 $$ -mapping method and the solution algorithm.

GPU-accelerated Bloch simulations and MR-STAT reconstructions using the Julia programming language.

PURPOSE: MR-STAT is a relatively new multiparametric quantitative MRI technique in which quantitative paramater maps are obtained by solving a large-scale nonlinear optimization problem. Managing reconstruction times is one of the main challenges of MR-STAT. In this work we leverage GPU hardware to reduce MR-STAT reconstruction times. A highly optimized, GPU-compatible Bloch simulation toolbox is developed as part of this work that can be utilized for other quantitative MRI techniques as well. METHODS: The Julia programming language was used to develop a flexible yet highly performant and GPU-compatible Bloch simulation toolbox called BlochSimulators.jl. The runtime performance of the toolbox is benchmarked against other Bloch simulation toolboxes. Furthermore, a (partially matrix-free) modification of a previously presented (matrix-free) MR-STAT reconstruction algorithm is proposed and implemented using the Julia language on GPU hardware. The proposed algorithm is combined with BlochSimulators.jl and the resulting MR-STAT reconstruction times on GPU hardware are compared to previously presented MR-STAT reconstruction times. RESULTS: The BlochSimulators.jl package demonstrates superior runtime performance on both CPU and GPU hardware when compared to other existing Bloch simulation toolboxes. The GPU-accelerated partially matrix-free MR-STAT reconstruction algorithm, which relies on BlochSimulators.jl, allows for reconstructions of 68 seconds per two-dimensional (2D slice). CONCLUSION: By combining the proposed Bloch simulation toolbox and the partially matrix-free reconstruction algorithm, 2D MR-STAT reconstructions can be performed in the order of one minute on a modern GPU card. The Bloch simulation toolbox can be utilized for other quantitative MRI techniques as well, for example for online dictionary generation for MR Fingerprinting.

Deep-learning-based joint rigid and deformable contour propagation for magnetic resonance imaging-guided prostate radiotherapy.

BACKGROUND: Deep learning-based unsupervised image registration has recently been proposed, promising fast registration. However, it has yet to be adopted in the online adaptive magnetic resonance imaging-guided radiotherapy (MRgRT) workflow. PURPOSE: In this paper, we design an unsupervised, joint rigid, and deformable registration framework for contour propagation in MRgRT of prostate cancer. METHODS: Three-dimensional pelvic T2-weighted MRIs of 143 prostate cancer patients undergoing radiotherapy were collected and divided into 110, 13, and 20 patients for training, validation, and testing. We designed a framework using convolutional neural networks (CNNs) for rigid and deformable registration. We selected the deformable registration network architecture among U-Net, MS-D Net, and LapIRN and optimized the training strategy (end-to-end vs. sequential). The framework was compared against an iterative baseline registration. We evaluated registration accuracy (the Dice and Hausdorff distance of the prostate and bladder contours), structural similarity index, and folding percentage to compare the methods. We also evaluated the framework's robustness to rigid and elastic deformations and bias field perturbations. RESULTS: The end-to-end trained framework comprising LapIRN for the deformable component achieved the best median (interquartile range) prostate and bladder Dice of 0.89 (0.85-0.91) and 0.86 (0.80-0.91), respectively. This accuracy was comparable to the iterative baseline registration: prostate and bladder Dice of 0.91 (0.88-0.93) and 0.86 (0.80-0.92). The best models complete rigid and deformable registration in 0.002 (0.0005) and 0.74 (0.43) s (Nvidia Tesla V100-PCIe 32 GB GPU), respectively. We found that the models are robust to translations up to 52 mm, rotations up to 15 ∘ $^\circ$ , elastic deformations up to 40 mm, and bias fields. CONCLUSIONS: Our proposed unsupervised, deep learning-based registration framework can perform rigid and deformable registration in less than a second with contour propagation accuracy comparable with iterative registration.

High SNR full brain relaxometry at 7T by accelerated MR-STAT.

PURPOSE: To demonstrate the feasibility and robustness of the Magnetic Resonance Spin TomogrAphy in Time-domain (MR-STAT) framework for fast, high SNR relaxometry at 7T. METHODS: To deploy MR-STAT on 7T-systems, we designed optimized flip-angles using the BLAKJac-framework that incorporates the SAR-constraints. Transmit RF-inhomogeneities were mitigated by including a measured B 1 + $$ {B}_1^{+} $$ -map in the reconstruction. Experiments were performed on a gel-phantom and on five volunteers to explore the robustness of the sequence and its sensitivity to B 1 + $$ {B}_1^{+} $$ inhomogeneities. The SNR-gain at 7T was explored by comparing phantom and in vivo results to MR-STAT at 3T in terms of SNR-efficiency. RESULTS: The higher SNR at 7T enabled two-fold acceleration with respect to current 2D MR-STAT protocols at lower field strengths. The resulting scan had whole-brain coverage, with 1 x 1 x 3 mm3 resolution (1.5 mm slice-gap) and was acquired within 3 min including the B 1 + $$ {B}_1^{+} $$ -mapping. After B 1 + $$ {B}_1^{+} $$ -correction, the estimated T1 and T2 in a phantom showed a mean relative error of, respectively, 1.7% and 4.4%. In vivo, the estimated T1 and T2 in gray and white matter corresponded to the range of values reported in literature with a variation over the subjects of 1.0%-2.1% (WM-GM) for T1 and 4.3%-5.3% (WM-GM) for T2. We measured a higher SNR-efficiency at 7T (R = 2) than at 3T for both T1 and T2 with, respectively, a 4.1 and 2.3 times increase in SNR-efficiency. CONCLUSION: We presented an accelerated version of MR-STAT tailored to high field (7T) MRI using a low-SAR flip-angle train and showed high quality parameter maps with an increased SNR-efficiency compared to MR-STAT at 3T.

Automated pelvic MRI measurements associated with urinary incontinence for prostate cancer patients undergoing radical prostatectomy.

BACKGROUND: Pelvic morphological parameters on magnetic resonance imaging (MRI), such as the membranous urethral length (MUL), can predict urinary incontinence after radical prostatectomy but are prone to interobserver disagreement. Our objective was to improve interobserver agreement among radiologists in measuring pelvic parameters using deep learning (DL)-based segmentation of pelvic structures on MRI scans. METHODS: Preoperative MRI was collected from 167 prostate cancer patients undergoing radical prostatectomy within our regional multicentric cohort. Two DL networks (nnU-Net) were trained on coronal and sagittal scans and evaluated on a test cohort using an 80/20% train-test split. Pelvic parameters were manually measured by three abdominal radiologists on raw MRI images and with the use of DL-generated segmentations. Automated measurements were also performed for the pelvic parameters. Interobserver agreement was evaluated using the intraclass correlation coefficient (ICC) and the Bland-Altman plot. RESULTS: The DL models achieved median Dice similarity coefficient (DSC) values of 0.85-0.97 for coronal structures and 0.87-0.98 for sagittal structures. When radiologists used DL-generated segmentations of pelvic structures, the interobserver agreement for sagittal MUL improved from 0.64 (95% confidence interval 0.28-0.83) to 0.91 (95% CI 0.84-0.95). Furthermore, there was an increase in ICC values for the obturator internus muscle from 0.74 (95% CI 0.42-0.87) to 0.86 (95% CI 0.75-0.92) and for the levator ani muscle from 0.40 (95% CI 0.05-0.66) to 0.61 (95% CI 0.31-0.78). CONCLUSIONS: DL-based automated segmentation of pelvic structures improved interobserver agreement in measuring pelvic parameters on preoperative MRI scans. RELEVANCE STATEMENT: The implementation of deep learning segmentations allows for more consistent measurements of pelvic parameters by radiologists. Standardized measurements are crucial for incorporating these parameters into urinary continence prediction models. KEY POINTS: • DL-generated segmentations improve interobserver agreement for pelvic measurements among radiologists. • Membranous urethral length measurement improved from substantial to almost perfect agreement. • Artificial intelligence enhances objective pelvic parameter assessment for continence prediction models.

A three-dimensional Magnetic Resonance Spin Tomography in Time-domain protocol for high-resolution multiparametric quantitative magnetic resonance imaging.

Magnetic Resonance Spin TomogrAphy in Time-domain (MR-STAT) is a multiparametric quantitative MR framework, which allows for simultaneously acquiring quantitative tissue parameters such as T1, T2, and proton density from one single short scan. A typical two-dimensional (2D) MR-STAT acquisition uses a gradient-spoiled, gradient-echo sequence with a slowly varying RF flip-angle train and Cartesian readouts, and the quantitative tissue maps are reconstructed by an iterative, model-based optimization algorithm. In this work, we design a three-dimensional (3D) MR-STAT framework based on previous 2D work, in order to achieve better image signal-to-noise ratio, higher though-plane resolution, and better tissue characterization. Specifically, we design a 7-min, high-resolution 3D MR-STAT sequence, and the corresponding two-step reconstruction algorithm for the large-scale dataset. To reduce the long acquisition time, Cartesian undersampling strategies such as SENSE are adopted in our transient-state quantitative framework. To reduce the computational burden, a data-splitting scheme is designed for decoupling the 3D reconstruction problem into independent 2D reconstructions. The proposed 3D framework is validated by numerical simulations, phantom experiments, and in vivo experiments. High-quality knee quantitative maps with 0.8 × 0.8 × 1.5 mm3 resolution and bilateral lower leg maps with 1.6 mm isotropic resolution can be acquired using the proposed 7-min acquisition sequence and the 3-min-per-slice decoupled reconstruction algorithm. The proposed 3D MR-STAT framework could have wide clinical applications in the future.

Water diffusion and T2 quantification in transient-state MRI: the effect of RF pulse sequence.

In quantitative measurement of the T 2 value of tissues, the diffusion of water molecules has been recognized as a confounder. This is most notably so for transient-state quantitative mapping techniques, which allow simultaneous estimation of T 1 and T 2 . In prior work, apparently conflicting conclusions are presented on the level of diffusion-induced bias on the T2 estimate. So far there is a lack of studies on the effect of the RF pulse angle sequence on the level of diffusion-induced bias. In this work, we show that the specific transient-state RF pulse sequence has a large effect on this level of bias. In particular, the bias level is strongly influenced by the mean value of the RF pulse angles. Also, for realistic values of the spoiling gradient area, we infer that the diffusion-induced bias is negligible for non-liquid human tissues; yet, for phantoms, the effect can be substantial (15% of the true T 2 value) for some RF pulse sequences. This should be taken into account in validation procedures.

Generalizable synthetic MRI with physics-informed convolutional networks.

BACKGROUND: Magnetic resonance imaging (MRI) provides state-of-the-art image quality for neuroimaging, consisting of multiple separately acquired contrasts. Synthetic MRI aims to accelerate examinations by synthesizing any desirable contrast from a single acquisition. PURPOSE: We developed a physics-informed deep learning-based method to synthesize multiple brain MRI contrasts from a single 5-min acquisition and investigate its ability to generalize to arbitrary contrasts. METHODS: A dataset of 55 subjects acquired with a clinical MRI protocol and a 5-min transient-state sequence was used. The model, based on a generative adversarial network, maps data acquired from the five-minute scan to "effective" quantitative parameter maps (q*-maps), feeding the generated PD, T1 , and T2 maps into a signal model to synthesize four clinical contrasts (proton density-weighted, T1 -weighted, T2 -weighted, and T2 -weighted fluid-attenuated inversion recovery), from which losses are computed. The synthetic contrasts are compared to an end-to-end deep learning-based method proposed by literature. The generalizability of the proposed method is investigated for five volunteers by synthesizing three contrasts unseen during training and comparing these to ground truth acquisitions via qualitative assessment and contrast-to-noise ratio (CNR) assessment. RESULTS: The physics-informed method matched the quality of the end-to-end method for the four standard contrasts, with structural similarity metrics above 0.75 ± 0.08 (±std), peak signal-to-noise ratios above 22.4 ± 1.9, representing a portion of compact lesions comparable to standard MRI. Additionally, the physics-informed method enabled contrast adjustment, and similar signal contrast and comparable CNRs to the ground truth acquisitions for three sequences unseen during model training. CONCLUSIONS: The study demonstrated the feasibility of physics-informed, deep learning-based synthetic MRI to generate high-quality contrasts and generalize to contrasts beyond the training data. This technology has the potential to accelerate neuroimaging protocols.

The Development and External Validation of Artificial Intelligence-Driven MRI-Based Models to Improve Prediction of Lesion-Specific Extraprostatic Extension in Patients with Prostate Cancer.

Adequate detection of the histopathological extraprostatic extension (EPE) of prostate cancer (PCa) remains a challenge using conventional radiomics on 3 Tesla multiparametric magnetic resonance imaging (3T mpMRI). This study focuses on the assessment of artificial intelligence (AI)-driven models with innovative MRI radiomics in predicting EPE of prostate cancer (PCa) at a lesion-specific level. With a dataset encompassing 994 lesions from 794 PCa patients who underwent robot-assisted radical prostatectomy (RARP) at two Dutch hospitals, the study establishes and validates three classification models. The models were validated on an internal validation cohort of 162 lesions and an external validation cohort of 189 lesions in terms of discrimination, calibration, net benefit, and comparison to radiology reporting. Notably, the achieved AUCs ranged from 0.86 to 0.91 at the lesion-specific level, demonstrating the superior accuracy of the random forest model over conventional radiological reporting. At the external test cohort, the random forest model was the best-calibrated model and demonstrated a significantly higher accuracy compared to radiological reporting (83% vs. 67%, p = 0.02). In conclusion, an AI-powered model that includes both existing and novel MRI radiomics improves the detection of lesion-specific EPE in prostate cancer.

Accelerated respiratory-resolved 4D-MRI with separable spatio-temporal neural networks.

BACKGROUND: Respiratory-resolved four-dimensional magnetic resonance imaging (4D-MRI) provides essential motion information for accurate radiation treatments of mobile tumors. However, obtaining high-quality 4D-MRI suffers from long acquisition and reconstruction times. PURPOSE: To develop a deep learning architecture to quickly acquire and reconstruct high-quality 4D-MRI, enabling accurate motion quantification for MRI-guided radiotherapy (MRIgRT). METHODS: A small convolutional neural network called MODEST is proposed to reconstruct 4D-MRI by performing a spatial and temporal decomposition, omitting the need for 4D convolutions to use all the spatio-temporal information present in 4D-MRI. This network is trained on undersampled 4D-MRI after respiratory binning to reconstruct high-quality 4D-MRI obtained by compressed sensing reconstruction. The network is trained, validated, and tested on 4D-MRI of 28 lung cancer patients acquired with a T1-weighted golden-angle radial stack-of-stars (GA-SOS) sequence. The 4D-MRI of 18, 5, and 5 patients were used for training, validation, and testing. Network performances are evaluated on image quality measured by the structural similarity index (SSIM) and motion consistency by comparing the position of the lung-liver interface on undersampled 4D-MRI before and after respiratory binning. The network is compared to conventional architectures such as a U-Net, which has 30 times more trainable parameters. RESULTS: MODEST can reconstruct high-quality 4D-MRI with higher image quality than a U-Net, despite a thirty-fold reduction in trainable parameters. High-quality 4D-MRI can be obtained using MODEST in approximately 2.5 min, including acquisition, processing, and reconstruction. CONCLUSION: High-quality accelerated 4D-MRI can be obtained using MODEST, which is particularly interesting for MRIgRT.

Data-driven electrical conductivity brain imaging using 3 T MRI.

Magnetic resonance electrical properties tomography (MR-EPT) is a non-invasive measurement technique that derives the electrical properties (EPs, e.g., conductivity or permittivity) of tissues in the radiofrequency range (64 MHz for 1.5 T and 128 MHz for 3 T MR systems). Clinical studies have shown the potential of tissue conductivity as a biomarker. To date, model-based conductivity reconstructions rely on numerical assumptions and approximations, leading to inaccuracies in the reconstructed maps. To address such limitations, we propose an artificial neural network (ANN)-based non-linear conductivity estimator trained on simulated data for conductivity brain imaging. Network training was performed on 201 synthesized T2-weighted spin-echo (SE) data obtained from the finite-difference time-domain (FDTD) electromagnetic (EM) simulation. The dataset was composed of an approximated T2-w SE magnitude and transceive phase information. The proposed method was tested three in-silico and in-vivo on two volunteers and three patients' data. For comparison purposes, various conventional phase-based EPT reconstruction methods were used that ignore B 1 + magnitude information, such as Savitzky-Golay kernel combined with Gaussian filter (S-G Kernel), phase-based convection-reaction EPT (cr-EPT), magnitude-weighted polynomial-fitting phase-based EPT (Poly-Fit), and integral-based phase-based EPT (Integral-based). From the in-silico experiments, quantitative analysis showed that the proposed method provides more accurate and improved quality (e.g., high structural preservation) conductivity maps compared to conventional reconstruction methods. Representatively, in the healthy brain in-silico phantom experiment, the proposed method yielded mean conductivity values of 1.97 ± 0.20 S/m for CSF, 0.33 ± 0.04 S/m for WM, and 0.52 ± 0.08 S/m for GM, which were closer to the ground-truth conductivity (2.00, 0.30, 0.50 S/m) than the integral-based method (2.56 ± 2.31, 0.39 ± 0.12, 0.68 ± 0.33 S/m). In-vivo ANN-based conductivity reconstructions were also of improved quality compared to conventional reconstructions and demonstrated network generalizability and robustness to in-vivo data and pathologies. The reported in-vivo brain conductivity values were in agreement with literatures. In addition, the proposed method was observed for various SNR levels (SNR levels = 10, 20, 40, and 58) and repeatability conditions (the eight acquisitions with the number of signal averages = 1). The preliminary investigations on brain tumor patient datasets suggest that the network trained on simulated dataset can generalize to unforeseen in-vivo pathologies, thus demonstrating its potential for clinical applications.

Exploring contrast generalisation in deep learning-based brain MRI-to-CT synthesis.

BACKGROUND: Synthetic computed tomography (sCT) has been proposed and increasingly clinically adopted to enable magnetic resonance imaging (MRI)-based radiotherapy. Deep learning (DL) has recently demonstrated the ability to generate accurate sCT from fixed MRI acquisitions. However, MRI protocols may change over time or differ between centres resulting in low-quality sCT due to poor model generalisation. PURPOSE: investigating domain randomisation (DR) to increase the generalisation of a DL model for brain sCT generation. METHODS: CT and corresponding T1-weighted MRI with/without contrast, T2-weighted, and FLAIR MRI from 95 patients undergoing RT were collected, considering FLAIR the unseen sequence where to investigate generalisation. A "Baseline" generative adversarial network was trained with/without the FLAIR sequence to test how a model performs without DR. Image similarity and accuracy of sCT-based dose plans were assessed against CT to select the best-performing DR approach against the Baseline. RESULTS: The Baseline model had the poorest performance on FLAIR, with mean absolute error (MAE) = 106 ± 20.7 HU (mean ±σ). Performance on FLAIR significantly improved for the DR model with MAE = 99.0 ± 14.9 HU, but still inferior to the performance of the Baseline+FLAIR model (MAE = 72.6 ± 10.1 HU). Similarly, an improvement in γ-pass rate was obtained for DR vs Baseline. CONCLUSION: DR improved image similarity and dose accuracy on the unseen sequence compared to training only on acquired MRI. DR makes the model more robust, reducing the need for re-training when applying a model on sequences unseen and unavailable for retraining.

Deep learning for automated contouring of neurovascular structures on magnetic resonance imaging for prostate cancer patients.

Background and purpose: Manual contouring of neurovascular structures on prostate magnetic resonance imaging (MRI) is labor-intensive and prone to considerable interrater disagreement. Our aim is to contour neurovascular structures automatically on prostate MRI by deep learning (DL) to improve workflow and interrater agreement. Materials and methods: Segmentation of neurovascular structures was performed on pre-treatment 3.0 T MRI data of 131 prostate cancer patients (training [n = 105] and testing [n = 26]). The neurovascular structures include the penile bulb (PB), corpora cavernosa (CCs), internal pudendal arteries (IPAs), and neurovascular bundles (NVBs). Two DL networks, nnU-Net and DeepMedic, were trained for auto-contouring on prostate MRI and evaluated using volumetric Dice similarity coefficient (DSC), mean surface distances (MSD), Hausdorff distances, and surface DSC. Three radiation oncologists evaluated the DL-generated contours and performed corrections when necessary. Interrater agreement was assessed and the time required for manual correction was recorded. Results: nnU-Net achieved a median DSC of 0.92 (IQR: 0.90-0.93) for the PB, 0.90 (IQR: 0.86-0.92) for the CCs, 0.79 (IQR: 0.77-0.83) for the IPAs, and 0.77 (IQR: 0.72-0.81) for the NVBs, which outperformed DeepMedic for each structure (p < 0.03). nnU-Net showed a median MSD of 0.24 mm for the IPAs and 0.71 mm for the NVBs. The median interrater DSC ranged from 0.93 to 1.00, with the majority of cases (68.9%) requiring manual correction times under two minutes. Conclusions: DL enables reliable auto-contouring of neurovascular structures on pre-treatment MRI data, easing the clinical workflow in neurovascular-sparing MR-guided radiotherapy.

Real-time myocardial landmark tracking for MRI-guided cardiac radio-ablation using Gaussian Processes.

Objective.The high speed of cardiorespiratory motion introduces a unique challenge for cardiac stereotactic radio-ablation (STAR) treatments with the MR-linac. Such treatments require tracking myocardial landmarks with a maximum latency of 100 ms, which includes the acquisition of the required data. The aim of this study is to present a new method that allows to track myocardial landmarks from few readouts of MRI data, thereby achieving a latency sufficient for STAR treatments.Approach.We present a tracking framework that requires only few readouts of k-space data as input, which can be acquired at least an order of magnitude faster than MR-images. Combined with the real-time tracking speed of a probabilistic machine learning framework called Gaussian Processes, this allows to track myocardial landmarks with a sufficiently low latency for cardiac STAR guidance, including both the acquisition of required data, and the tracking inference.Main results.The framework is demonstrated in 2D on a motion phantom, andin vivoon volunteers and a ventricular tachycardia (arrhythmia) patient. Moreover, the feasibility of an extension to 3D was demonstrated byin silico3D experiments with a digital motion phantom. The framework was compared with template matching-a reference, image-based, method-and linear regression methods. Results indicate an order of magnitude lower total latency (<10 ms) for the proposed framework in comparison with alternative methods. The root-mean-square-distances and mean end-point-distance with the reference tracking method was less than 0.8 mm for all experiments, showing excellent (sub-voxel) agreement.Significance.The high accuracy in combination with a total latency of less than 10 ms-including data acquisition and processing-make the proposed method a suitable candidate for tracking during STAR treatments. Additionally, the probabilistic nature of the Gaussian Processes also gives access to real-time prediction uncertainties, which could prove useful for real-time quality assurance during treatments.

Gaussian Processes for real-time 3D motion and uncertainty estimation during MR-guided radiotherapy.

Respiratory motion during radiotherapy causes uncertainty in the tumor's location, which is typically addressed by an increased radiation area and a decreased dose. As a result, the treatments' efficacy is reduced. The recently proposed hybrid MR-linac scanner holds the promise to efficiently deal with such respiratory motion through real-time adaptive MR-guided radiotherapy (MRgRT). For MRgRT, motion-fields should be estimated from MR-data and the radiotherapy plan should be adapted in real-time according to the estimated motion-fields. All of this should be performed with a total latency of maximally 200 ms, including data acquisition and reconstruction. A measure of confidence in such estimated motion-fields is highly desirable, for instance to ensure the patient's safety in case of unexpected and undesirable motion. In this work, we propose a framework based on Gaussian Processes to infer 3D motion-fields and uncertainty maps in real-time from only three readouts of MR-data. We demonstrated an inference frame rate up to 69 Hz including data acquisition and reconstruction, thereby exploiting the limited amount of required MR-data. Additionally, we designed a rejection criterion based on the motion-field uncertainty maps to demonstrate the framework's potential for quality assurance. The framework was validated in silico and in vivo on healthy volunteer data (n=5) acquired using an MR-linac, thereby taking into account different breathing patterns and controlled bulk motion. Results indicate end-point-errors with a 75th percentile below 1 mm in silico, and a correct detection of erroneous motion estimates with the rejection criterion. Altogether, the results show the potential of the framework for application in real-time MR-guided radiotherapy with an MR-linac.

Multi-echo MR thermometry in the upper leg at 7 T using near-harmonic 2D reconstruction for initialization.

PURPOSE: The aim of this work is the development of a thermometry method to measure temperature increases in vivo, with a precision and accuracy sufficient for validation against thermal simulations. Such an MR thermometry model would be a valuable tool to get an indication on one of the major safety concerns in MR imaging: the tissue heating occurring due to radiofrequency (RF) exposure. To prevent excessive temperature rise, RF power deposition, expressed as specific absorption rate, cannot exceed predefined thresholds. Using these thresholds, MRI has demonstrated an extensive history of safe usage. Nevertheless, MR thermometry would be a valuable tool to address some of the unmet needs in the area of RF safety assessment, such as validation of specific absorption rate and thermal simulations, investigation of local peak temperatures during scanning, or temperature-based safety guidelines. METHODS: The harmonic initialized model-based multi-echo approach is proposed. The method combines a previously published model-based multi-echo water/fat separated approach with an also previously published near-harmonic 2D reconstruction method. The method is tested on the human thigh with a multi-transmit array at 7 T, in three volunteers, and for several RF shims. RESULTS: Precision and accuracy are improved considerably compared to a previous fat-referenced method (precision: 0.09 vs. 0.19°C). Comparison of measured temperature rise distributions to subject-specific simulated counterparts show good relative agreement for multiple RF shim settings. CONCLUSION: The high precision shows promising potential for validation purposes and other RF safety applications.