A novel approach to high resolution fetal brain MR imaging

High-Resolution 3D MRI With Deep Generative Networks via Novel Slice-Profile Transformation Super-Resolution

High-resolution magnetic resonance imaging (MRI) sequences, such as 3D turbo or fast spin-echo (TSE/FSE) imaging, are clinically desirable but suffer from long scanning time-related blurring when reformatted into preferred orientations. Instead, multi-slice two-dimensional (2D) TSE imaging is commonly used because of its high in-plane resolution but is limited clinically by poor through-plane resolution due to elongated voxels and the inability to generate multi-planar reformations due to staircase artifacts. Therefore, multiple 2D TSE scans are acquired in various orthogonal imaging planes, increasing the overall MRI scan time. In this study, we propose a novel slice-profile transformation super-resolution (SPTSR) framework with deep generative learning for through-plane super-resolution (SR) of multi-slice 2D TSE imaging. The deep generative networks were trained by synthesized low-resolution training input via slice-profile downsampling (SP-DS), and the trained networks inferred on the slice profile convolved (SP-conv) testing input for 5.5x through-plane SR. The network output was further slice-profile deconvolved (SP-deconv) to achieve an isotropic super-resolution. Compared to SMORE SR method and the networks trained by conventional downsampling, our SPTSR framework demonstrated the best overall image quality from 50 testing cases, evaluated by two abdominal radiologists. The quantitative analysis cross-validated the expert reader study results. 3D simulation experiments confirmed the quantitative improvement of the proposed SPTSR and the effectiveness of the SP-deconv step, compared to 3D ground-truths. Ablation studies were conducted on the individual contributions of SP-DS and SP-conv, networks structure, training dataset size, and different slice profiles.

Scan-Specific Generative Neural Network for MRI Super-Resolution Reconstruction

The interpretation and analysis of Magnetic resonance imaging (MRI) benefit from high spatial resolution. Unfortunately, direct acquisition of high spatial resolution MRI is time-consuming and costly, which increases the potential for motion artifact, and suffers from reduced signal-to-noise ratio (SNR). Super-resolution reconstruction (SRR) is one of the most widely used methods in MRI since it allows for the trade-off between high spatial resolution, high SNR, and reduced scan times. Deep learning has emerged for improved SRR as compared to conventional methods. However, current deep learning-based SRR methods require large-scale training datasets of high-resolution images, which are practically difficult to obtain at a suitable SNR. We sought to develop a methodology that allows for dataset-free deep learning-based SRR, through which to construct images with higher spatial resolution and of higher SNR than can be practically obtained by direct Fourier encoding. We developed a dataset-free learning method that leverages a generative neural network trained for each specific scan or set of scans, which in turn, allows for SRR tailored to the individual patient. With the SRR from three short duration scans, we achieved high quality brain MRI at an isotropic spatial resolution of 0.125 cubic mm with six minutes of imaging time for T2 contrast and an average increase of 7.2 dB (34.2%) in SNR to these short duration scans. Motion compensation was achieved by aligning the three short duration scans together. We assessed our technique on simulated MRI data and clinical data acquired from 15 subjects. Extensive experimental results demonstrate that our approach achieved superior results to state-of-the-art methods, while in parallel, performed at reduced cost as scans delivered with direct high-resolution acquisition.

Fast and High-Resolution Neonatal Brain MRI Through Super-Resolution Reconstruction From Acquisitions With Variable Slice Selection Direction

The brain of neonates is small in comparison to adults. Imaging at typical resolutions such as one cubic mm incurs more partial voluming artifacts in a neonate than in an adult. The interpretation and analysis of MRI of the neonatal brain benefit from a reduction in partial volume averaging that can be achieved with high spatial resolution. Unfortunately, direct acquisition of high spatial resolution MRI is slow, which increases the potential for motion artifact, and suffers from reduced signal-to-noise ratio. The purpose of this study is thus that using super-resolution reconstruction in conjunction with fast imaging protocols to construct neonatal brain MRI images at a suitable signal-to-noise ratio and with higher spatial resolution than can be practically obtained by direct Fourier encoding. We achieved high quality brain MRI at a spatial resolution of isotropic 0.4 mm with 6 min of imaging time, using super-resolution reconstruction from three short duration scans with variable directions of slice selection. Motion compensation was achieved by aligning the three short duration scans together. We applied this technique to 20 newborns and assessed the quality of the images we reconstructed. Experiments show that our approach to super-resolution reconstruction achieved considerable improvement in spatial resolution and signal-to-noise ratio, while, in parallel, substantially reduced scan times, as compared to direct high-resolution acquisitions. The experimental results demonstrate that our approach allowed for fast and high-quality neonatal brain MRI for both scientific research and clinical studies.

Autoencoder-Inspired Convolutional Network-Based Super-Resolution Method in MRI

OBJECTIVE

To introduce an MRI in-plane resolution enhancement method that estimates High-Resolution (HR) MRIs from Low-Resolution (LR) MRIs.

METHOD & MATERIALS

Previous CNN-based MRI super-resolution methods cause loss of input image information due to the pooling layer. An Autoencoder-inspired Convolutional Network-based Super-resolution (ACNS) method was developed with the deconvolution layer that extrapolates the missing spatial information by the convolutional neural network-based nonlinear mapping between LR and HR features of MRI. Simulation experiments were conducted with virtual phantom images and thoracic MRIs from four volunteers. The Peak Signal-to-Noise Ratio (PSNR), Structure SIMilarity index (SSIM), Information Fidelity Criterion (IFC), and computational time were compared among: ACNS; Super-Resolution Convolutional Neural Network (SRCNN); Fast Super-Resolution Convolutional Neural Network (FSRCNN); Deeply-Recursive Convolutional Network (DRCN).

RESULTS

ACNS achieved comparable PSNR, SSIM, and IFC results to SRCNN, FSRCNN, and DRCN. However, the average computation speed of ACNS was 6, 4, and 35 times faster than SRCNN, FSRCNN, and DRCN, respectively under the computer setup used with the actual average computation time of 0.15 s per [Formula: see text].

A Sparse Volume Reconstruction Method for Fetal Brain MRI Using Adaptive Kernel Regression

Slice-to-volume reconstruction (SVR) method can deal well with motion artifacts and provide high-quality 3D image data for fetal brain MRI. However, the problem of sparse sampling is not well addressed in the SVR method. In this paper, we mainly focus on the sparse volume reconstruction of fetal brain MRI from multiple stacks corrupted with motion artifacts. Based on the SVR framework, our approach includes the slice-to-volume 2D/3D registration, the point spread function- (PSF-) based volume update, and the adaptive kernel regression-based volume update. The adaptive kernel regression can deal well with the sparse sampling data and enhance the detailed preservation by capturing the local structure through covariance matrix. Experimental results performed on clinical data show that kernel regression results in statistical improvement of image quality for sparse sampling data with the parameter setting of the structure sensitivity 0.4, the steering kernel size of 7 × 7 × 7 and steering smoothing bandwidth of 0.5. The computational performance of the proposed GPU-based method can be over 90 times faster than that on CPU.

A discriminative feature selection approach for shape analysis: Application to fetal brain cortical folding

The development of post-processing reconstruction techniques has opened new possibilities for the study of in-utero fetal brain MRI data. Recent cortical surface analysis have led to the computation of quantitative maps characterizing brain folding of the developing brain. In this paper, we describe a novel feature selection-based approach that is used to extract the most discriminative and sparse set of features of a given dataset. The proposed method is used to sparsely characterize cortical folding patterns of an in-utero fetal MR dataset, labeled with heterogeneous gestational age ranging from 26 weeks to 34 weeks. The proposed algorithm is validated on a synthetic dataset with both linear and non-linear dynamics, supporting its ability to capture deformation patterns across the dataset within only a few features. Results on the fetal brain dataset show that the temporal process of cortical folding related to brain maturation can be characterized by a very small set of points, located in anatomical regions changing across time. Quantitative measurements of growth against time are extracted from the set selected features to compare multiple brain regions (e.g. lobes and hemispheres) during the considered period of gestation.

Motion correction in MRI of the brain

Subject motion in MRI is a relevant problem in the daily clinical routine as well as in scientific studies. Since the beginning of clinical use of MRI, many research groups have developed methods to suppress or correct motion artefacts. This review focuses on rigid body motion correction of head and brain MRI and its application in diagnosis and research. It explains the sources and types of motion and related artefacts, classifies and describes existing techniques for motion detection, compensation and correction and lists established and experimental approaches. Retrospective motion correction modifies the MR image data during the reconstruction, while prospective motion correction performs an adaptive update of the data acquisition. Differences, benefits and drawbacks of different motion correction methods are discussed.

Fast Volume Reconstruction From Motion Corrupted Stacks of 2D Slices

Capturing an enclosing volume of moving subjects and organs using fast individual image slice acquisition has shown promise in dealing with motion artefacts. Motion between slice acquisitions results in spatial inconsistencies that can be resolved by slice-to-volume reconstruction (SVR) methods to provide high quality 3D image data. Existing algorithms are, however, typically very slow, specialised to specific applications and rely on approximations, which impedes their potential clinical use. In this paper, we present a fast multi-GPU accelerated framework for slice-to-volume reconstruction. It is based on optimised 2D/3D registration, super-resolution with automatic outlier rejection and an additional (optional) intensity bias correction. We introduce a novel and fully automatic procedure for selecting the image stack with least motion to serve as an initial registration target. We evaluate the proposed method using artificial motion corrupted phantom data as well as clinical data, including tracked freehand ultrasound of the liver and fetal Magnetic Resonance Imaging. We achieve speed-up factors greater than 30 compared to a single CPU system and greater than 10 compared to currently available state-of-the-art multi-core CPU methods. We ensure high reconstruction accuracy by exact computation of the point-spread function for every input data point, which has not previously been possible due to computational limitations. Our framework and its implementation is scalable for available computational infrastructures and tests show a speed-up factor of 1.70 for each additional GPU. This paves the way for the online application of image based reconstruction methods during clinical examinations. The source code for the proposed approach is publicly available.

MRI evaluation and safety in the developing brain

Magnetic resonance imaging (MRI) evaluation of the developing brain has dramatically increased over the last decade. Faster acquisitions and the development of advanced MRI sequences, such as magnetic resonance spectroscopy (MRS), diffusion tensor imaging (DTI), perfusion imaging, functional MR imaging (fMRI), and susceptibility-weighted imaging (SWI), as well as the use of higher magnetic field strengths has made MRI an invaluable tool for detailed evaluation of the developing brain. This article will provide an overview of the use and challenges associated with 1.5-T and 3-T static magnetic fields for evaluation of the developing brain. This review will also summarize the advantages, clinical challenges, and safety concerns specifically related to MRI in the fetus and newborn, including the implications of increased magnetic field strength, logistics related to transporting and monitoring of neonates during scanning, and sedation considerations, and a discussion of current technologies such as MRI conditional neonatal incubators and dedicated small-foot print neonatal intensive care unit (NICU) scanners.

Motion correction of magnetic resonance imaging data by using adaptive moving least squares method

Image artifacts caused by subject motion during the imaging sequence are one of the most common problems in magnetic resonance imaging (MRI) and often degrade the image quality. In this study, we develop a motion correction algorithm for the interleaved-MR acquisition. An advantage of the proposed method is that it does not require either additional equipment or redundant over-sampling. The general framework of this study is similar to that of Rohlfing et al. [1], except for the introduction of the following fundamental modification. The three-dimensional (3-D) scattered data approximation method is used to correct the artifacted data as a post-processing step. In order to obtain a better match to the local structures of the given image, we use the data-adapted moving least squares (MLS) method that can improve the performance of the classical method. Numerical results are provided to demonstrate the advantages of the proposed algorithm.

Fast, free-breathing, in vivo fetal imaging using time-resolved 3D MRI technique: preliminary results

Fetal MR imaging is very challenging due to the movement of fetus and the breathing motion of the mother. Current clinical protocols involve quick 2D scouting scans to determine scan plane and often several attempts to reorient the scan plane when the fetus moves. This makes acquisition of fetal MR images clinically challenging and results in long scan times in order to obtain images that are of diagnostic quality. Compared to 2D imaging, 3D imaging of the fetus has many advantages such as higher SNR and ability to reformat images in multiple planes. However, it is more sensitive to motion and challenging for fetal imaging due to irregular fetal motion in addition to maternal breathing and cardiac motion. This aim of this study is to develop a fast 3D fetal imaging technique to resolve the challenge of imaging the moving fetus. This 3D imaging sequence has multi-echo radial sampling in-plane and conventional Cartesian encoding through plane, which provides motion robustness and high data acquisition efficiency. The utilization of a golden-ratio based projection profile allows flexible time-resolved image reconstruction with arbitrary temporal resolution at arbitrary time points as well as high signal-to-noise and contrast-to-noise ratio. The nice features of the developed image technique allow the 3D visualization of the movements occurring throughout the scan. In this study, we applied this technique to three human subjects for fetal MRI and achieved promising preliminary results of fetal brain, heart and lung imaging.

Efficient total variation algorithm for fetal brain MRI reconstruction

Fetal MRI reconstruction aims at finding a high-resolution image given a small set of low-resolution images. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has considered several regularization terms s.a. Dirichlet/Laplacian energy, Total Variation (TV)- based energies and more recently non-local means. Although TV energies are quite attractive because of their ability in edge preservation, standard explicit steepest gradient techniques have been applied to optimize fetal-based TV energies. The main contribution of this work lies in the introduction of a well-posed TV algorithm from the point of view of convex optimization. Specifically, our proposed TV optimization algorithm or fetal reconstruction is optimal w.r.t. the asymptotic and iterative convergence speeds O(1/n2) and O(1/√ε), while existing techniques are in O(1/n2) and O(1/√ε). We apply our algorithm to (1) clinical newborn data, considered as ground truth, and (2) clinical fetal acquisitions. Our algorithm compares favorably with the literature in terms of speed and accuracy.

Advanced MR Imaging Technologies in Fetuses

Fetal Magnetic Resonance Imaging (MRI) on clinical scanners has increasingly been realized as a powerful imaging tool and applied for studying the brain abnormalities and the potential of neurodevelopmental disabilities in vivo. The primarily used multi-echo fast imaging sequences reduce the motion artifacts with a tradeoff of image Signal-to-Noise Ratio (SNR) and resolution. In Radio Frequency (RF) hardware for MR signal excitation and reception, there are lack of dedicated RF coils for fetal imaging providing optimized performance in acquisition and safety. There is an urgent demand for novel hardware and fast imaging technology developments to overcome motion artifacts and improve sensitivity and safety. Recent studies have demonstrated that dedicated fetal RF transceiver arrays can improve the SNR, image coverage, and safety. In addition, emerging fast imaging technologies such as parallel imaging and compressed sensing would be advantageous in improving imaging speed and thus reducing motion artifacts in fetal imaging.

White matter maturation of normal human fetal brain. An in vivo diffusion tensor tractography study

We demonstrate for the first time the ability to determine in vivo and in utero the transitions between the main stages of white matter (WM) maturation in normal human fetuses using magnetic resonance diffusion tensor imaging (DTI) tractography. Biophysical characteristics of water motion are used as an indirect probe to evaluate progression of the tissue matrix organization in cortico-spinal tracts (CSTs), optic radiations (OR), and corpus callosum (CC) in 17 normal human fetuses explored between 23 and 38 weeks of gestation (GW) and selected strictly on minimal motion artifacts. Nonlinear polynomial (third order) curve fittings of normalized longitudinal and radial water diffusivities (Z-scores) as a function of age identify three different phases of maturation with specific dynamics for each WM bundle type. These phases may correspond to distinct cellular events such as axonal organization, myelination gliosis, and myelination, previously reported by other groups on post-mortem fetuses using immunostaining methods. According to the DTI parameter dynamics, we suggest that myelination (phase 3) appears early in the CSTs, followed by the OR and by the CC, respectively. DTI tractography provides access to a better understanding of fetal WM maturation.

Fetal hippocampal development: analysis by magnetic resonance imaging volumetry

The hippocampal formation plays an important role in learning and memory; however, data on its development in utero in humans are limited. This study was performed to evaluate hippocampal development in healthy fetuses using 3D reconstructed MRI. A cohort of 20 healthy pregnant women underwent prenatal MRI at a median GA of 24.9 wk (range, 21.3-31.9 wk); six of the women also had a second fetal MRI performed at a 6-wk interval. Routine 2D ultrafast T2-weighted images were used to reconstruct a 3D volume image, which was then used to manually segment the right and left hippocampi. Total hippocampal volume was calculated for each subject and compared against GA. There was a linear increase in total hippocampal volume with increasing GA (p < 0.001). For subjects scanned twice, there was an increase in hippocampal size on the second fetal MRI (p = 0.0004). This represents the first volumetric study of fetal hippocampal development in vivo. This normative volumetric data will be helpful for future comparison studies of suspected developmental abnormalities of hippocampal structure and function.

Atlas-based segmentation of developing tissues in the human brain with quantitative validation in young fetuses

Imaging of the human fetus using magnetic resonance (MR) is an essential tool for quantitative studies of normal as well as abnormal brain development in utero. However, because of fundamental differences in tissue types, tissue properties and tissue distribution between the fetal and adult brain, automated tissue segmentation techniques developed for adult brain anatomy are unsuitable for this data. In this paper, we describe methodology for automatic atlas-based segmentation of individual tissue types in motion-corrected 3D volumes reconstructed from clinical MR scans of the fetal brain. To generate anatomically correct automatic segmentations, we create a set of accurate manual delineations and build an in utero 3D statistical atlas of tissue distribution incorporating developing gray and white matter as well as transient tissue types such as the germinal matrix. The probabilistic atlas is associated with an unbiased average shape and intensity template for registration of new subject images to the space of the atlas. Quantitative whole brain 3D validation of tissue labeling performed on a set of 14 fetal MR scans (20.57-22.86 weeks gestational age) demonstrates that this atlas-based EM segmentation approach achieves consistently high DSC performance for the main tissue types in the fetal brain. This work indicates that reliable measures of brain development can be automatically derived from clinical MR imaging and opens up possibility of further 3D volumetric and morphometric studies with multiple fetal subjects.

MR imaging of the fetal brain

Fetal MRI is clinically performed to evaluate the brain in cases where an abnormality is detected by prenatal sonography. These most commonly include ventriculomegaly, abnormalities of the corpus callosum, and abnormalities of the posterior fossa. Fetal MRI is also increasingly performed to evaluate fetuses who have normal brain findings on prenatal sonogram but who are at increased risk for neurodevelopmental abnormalities, such as complicated monochorionic twin pregnancies. This paper will briefly discuss the common clinical conditions imaged by fetal MRI as well as recent advances in fetal MRI research.

Volume reconstruction by inverse interpolation: application to interleaved MR motion correction

We introduce in this work a novel algorithm for volume reconstruction from data acquired on an irregular grid, e.g., from multiple co-registered images. The algorithm, which is based on an inverse interpolation formalism, is superior to other methods in particular when the input images have lower spatial resolution than the reconstructed image. Local intensity bounds are enforced by an L-BFGSB optimizer, regularize the reconstruction problem, and preserve the intensity distribution of the input images. We demonstrate the usefulness of our method by applying it to retrospective motion correction in interleaved MR images.

In-utero three dimension high resolution fetal brain diffusion tensor imaging

We present a methodology to achieve 3D high resolution in-utero fetal brain DTI that shows excellent ADC as well as promising FA maps. After continuous DTI scanning to acquire a repeated series of parallel slices with 15 diffusion directions, image registration is used to realign the images to correct for fetal motion. Once aligned, the diffusion images are treated as irregularly sampled data where each voxel is associated with an appropriately rotated diffusion direction, and used to estimate the diffusion tensor on a regular grid. The method has been tested successful on eight fetuses and has been validated on adults imaged at 1.5T.