Resolution Magnetic Resonance Imaging Research Articles

Quantum-enhanced MRI Sensitivity: Dissolution-dynamic Nuclear and Parahydrogen-induced Polarization

Contemporarily, magnetic resonance imaging (MRI) has been widely utilized in medical diagnostic. Among various features, the signal intensities serve as the key role in resolution of the detection results. In general, the MRI signal intensities can be substantially increased by several orders of magnitude via dissolution-dynamic nuclear polarization (d-DNP) and parahydrogen-induced polarization (PHIP). This study exhibits the general principles and components of the preparation of these two methods, as well as illustrates the current applications and limitations of d-DNP and PHIP both theoretically and analytically. The experimental conditions, including temperature, magnetic field strength, and whether microwave irradiation is required, are described and contrasted. According to the analysis, the advantages and drawbacks of these two approaches have been compared, along with expectations and outlooks of the future development of quantum-enhanced MRI techniques, in terms of the MRI signal sensitivity. Overall, these results shed light on guiding further exploration of enhancing the resolution of MRI.

Visualizing cell–cell communication using synthetic notch activated MRI

Cell-cell communication plays a fundamental role in multicellular organisms. Cell-based cancer immunotherapies rely on the ability of innate or engineered receptors on immune cells to engage specific antigens on cancer cells to induce tumor kill. Toimprove the development and translation of these therapies, imaging tools capable of noninvasively and spatiotemporally visualizing immune-cancer cell interactions would be highly valuable. Using the synthetic Notch (SynNotch) system, we engineered T cells that upon interaction with a chosen antigen (CD19) on neighboring cancer cells induce the expression of optical reporter genes and the human-derived, magnetic resonance imaging (MRI) reporter gene organic anion transporting polypeptide 1B3 (OATP1B3). Administration of engineered T cells induced the antigen-dependent expression of all our reporter genes in mice bearing CD19-positive tumors but not CD19-negative tumors. Notably, due to the high spatial resolution and tomographic nature of MRI, contrast-enhanced foci within CD19-positive tumors representing OATP1B3-expressing T cells were clearly visible and their distribution was readily mapped. We then extended this technology onto human natural killer-92 (NK-92) cells, observing similar CD19-dependent reporter activity in tumor-bearing mice. Furthermore, we show that when delivered intravenously, engineered NK-92 cells can be detected via bioluminescence imaging in a systemic cancer model. With continued work, this highly modular imaging strategy could aid in the monitoring of cell therapies in patients and, beyond this, augment our understanding of how different cell populations interact within the body during normal physiology or disease.

Radiolabeled Iron Oxide Nanoparticles as Dual Modality Contrast Agents in SPECT/MRI and PET/MRI.

During the last decades, the utilization of imaging modalities such as single photon emission computed tomography (SPECT), positron emission tomography (PET), and magnetic resonance imaging (MRI) in every day clinical practice has enabled clinicians to diagnose diseases accurately at early stages. Radiolabeled iron oxide nanoparticles (RIONs) combine their intrinsic magnetic behavior with the extrinsic character of the radionuclide additive, so that they constitute a platform of multifaceted physical properties. Thus, at a practical level, RIONs serve as the physical parent of the so-called dual-modality contrast agents (DMCAs) utilized in SPECT/MRI and PET/MRI applications due to their ability to combine, at real time, the high sensitivity of SPECT or PET together with the high spatial resolution of MRI. This review focuses on the synthesis and in vivo investigation of both biodistribution and imaging efficacy of RIONs as potential SPECT/MRI or PET/MRI DMCAs.

Standardizing Brain Magnetic Resonance Imaging using Generative Adversarial Networks: A Multisite Study Approach

ABSTRACT Background: Magnetic resonance imaging (MRI) intensities vary across sites due to differences in acquisition protocols and hardware. Resolution also differs across centers. This hampers developing multisite deep learning models on MRI data. Objective: To standardize MRI intensities and resolution to enable multisite deep learning. Materials and Methods: T2-weighted brain MRI from 500 subjects across sites were split into training, validation and test sets. A generative adversarial network (GAN) model was developed to convert 64x64 low-resolution inputs to 256x256 standardized outputs. Preprocessing involved skull stripping, interpolation and intensity scaling. The generator used convolutional layers and residual blocks. Discriminator classified real/fake images. VGG perceptual loss was incorporated along with MSE and adversarial losses. Results: The GAN model achieved a structural similarity index of 0.9937 and feature similarity of 0.00122 versus ground truth. Intensity distribution was retained. The proposed pipeline reduced interpolation noise by 94% in extracted features. Conclusion: The proposed GAN pipeline can effectively standardize multisite brain MRI for intensity and resolution. By enabling multi-center data harmonization, this approach facilitates developing deep learning models through federated learning on MRI big data.

Development of a Super-Resolution Scheme for Pediatric Magnetic Resonance Brain Imaging Through Convolutional Neural Networks

Pediatric medical imaging represents a real challenge for physicians, as children who are patients often move during the examination, and it causes the appearance of different artifacts in the images. Thus, it is not possible to obtain good quality images for this target population limiting the possibility of evaluation and diagnosis in certain pathological conditions. Specifically, magnetic resonance imaging (MRI) is a technique that requires long acquisition times and, therefore, demands the use of sedation or general anesthesia to avoid the movement of the patient, which is really damaging in this specific population. Because ALARA (as low as reasonably achievable) principles should be considered for all imaging studies, one of the most important reasons for establishing novel MRI imaging protocols is to avoid the harmful effects of anesthesia/sedation. In this context, ground-breaking concepts and novel technologies, such as artificial intelligence, can help to find a solution to these challenges while helping in the search for underlying disease mechanisms. The use of new MRI protocols and new image acquisition and/or pre-processing techniques can aid in the development of neuroimaging studies for children evaluation, and their translation to pediatric populations. In this paper, a novel super-resolution method based on a convolutional neural network (CNN) in two and three dimensions to automatically increase the resolution of pediatric brain MRI acquired in a reduced time scheme is proposed. Low resolution images have been generated from an original high resolution dataset and used as the input of the CNN, while several scaling factors have been assessed separately. Apart from a healthy dataset, we also tested our model with pathological pediatric MRI, and it successfully recovers the original image quality in both visual and quantitative ways, even for available examples of dysplasia lesions. We hope then to establish the basis for developing an innovative free-sedation protocol in pediatric anatomical MRI acquisition.

The Role of Preoperative Magnetic Resonance Imaging in Assessing Neurovascular Compression Before Microvascular Decompression in Trigeminal Neuralgia

Preoperative magnetic resonance imaging (MRI) is a standard component of the preoperative clinical workup for patients before microvascular decompression (MVD). However, its ability to accurately exclude neurovascular compression of the trigeminal nerve is not well understood. We retrospectively reviewed 1020 patients with available preoperative MRI data before microvascular decompression. General patient demographics and clinical characteristics were collected for each case. We recorded both evidence of neurovascular conflict on preoperative MRI radiology notes and intraoperative compression from operative notes. Sensitivity, specificity, positive predictive value, and negative predictive value were determined for general MRI, high-resolution MRI, and non-high resolution. Overall, preoperative MRI before MVD showed a sensitivity of 75.8%, specificity of 65.8%, positive predictive value of 92.4%, and negative predictive value of 33.3% in predicting neurovascular compression of the trigeminal nerve. In particular, MRI was unable to identify 21.0% cases of sole arterial compression, 42.5% cases of sole venous compression, and combined arterial and venous compression in 18.5% of cases. A total of 958 patients (93.9%) underwent high-resolution preoperative MRI with skull base sequences. This imaging showed a sensitivity of 75.6%, specificity of 66.9%, positive predictive value of 92.5% and a negative predictive value of 33.4% in predicting trigeminal nerve neurovascular compression. Non-high-resolution MRI showed a sensitivity of 78.8%, specificity of 50.0%, positive predictive value of 89.1%, and negative predictive value of 31.3%. The negative predictive values of general, high-resolution, and non-high-resolution MRIs were all <50%. Preoperative MRI may offer a high predictive value for neurovascular conflict and should be part of the standard preoperative care workup for patients with trigeminal neuralgia. However, lack of neurovascular conflict on preoperative imaging is not sufficient to exclude patients from undergoing MVD.

Accurate quantification of blood flow wall shear stress using simulation-based imaging: a synthetic, comparative study

Simulation-based imaging (SBI) is a blood flow imaging technique that optimally fits a computational fluid dynamics (CFD) simulation to low-resolution, noisy magnetic resonance (MR) flow data to produce a high-resolution velocity field. In this work, we study the effectivity of SBI in predicting wall shear stress (WSS) relative to standard magnetic resonance imaging (MRI) postprocessing techniques using two synthetic numerical experiments: steady flow through an idealized, two-dimensional stenotic vessel and a model of an adult aorta. In particular, we study the sensitivity of these two approaches with respect to the Reynolds number of the underlying flow, the resolution of the MRI data, and the noise in the MRI data. We found that the SBI WSS reconstruction: (1) is insensitive to Reynolds number over the range considered (mathrm {Re} le 1000), (2) improves as the amount of MRI data increases and provides accurate reconstructions with as little as three MRI voxels per diameter, and (3) degrades linearly as the noise in the data increases with a slope determined by the resolution of the MRI data. We also consider the sensitivity of SBI to the resolution of the CFD mesh and found there is flexibility in the mesh used for SBI, although the WSS reconstruction becomes more sensitive to other parameters, particularly the resolution of the MRI data, for coarser meshes. This indicates a fundamental trade-off between scan time (i.e., MRI data quality and resolution) and reconstruction time using SBI, which is inherently different than the trade-off between scan time and reconstruction quality observed in standard MRI postprocessing techniques.

The effects of prenatal bisphenol A exposure on brain volume of children and young mice

Bisphenol A (BPA) is a synthetic chemical used for the manufacturing of plastics, epoxy resin, and many personal care products. This ubiquitous endocrine disruptor is detectable in the urine of over 80% of North Americans. Although adverse neurodevelopmental outcomes have been observed in children with high gestational exposure to BPA, the effects of prenatal BPA on brain structure remain unclear. Here, using magnetic resonance imaging (MRI), we studied the associations of maternal BPA exposure with children's brain structure, as well as the impact of comparable BPA levels in a mouse model. Our human data showed that most maternal BPA exposure effects on brain volumes were small, with the largest effects observed in the opercular region of the inferior frontal gyrus (ρ = −0.2754), superior occipital gyrus (ρ = −0.2556), and postcentral gyrus (ρ = 0.2384). In mice, gestational exposure to an equivalent level of BPA (2.25 μg BPA/kg bw/day) induced structural alterations in brain regions including the superior olivary complex (SOC) and bed nucleus of stria terminalis (BNST) with larger effect sizes (1.07≤ Cohens d ≤ 1.53). Human (n = 87) and rodent (n = 8 each group) sample sizes, while small, are considered adequate to perform the primary endpoint analysis. Combined, these human and mouse data suggest that gestational exposure to low levels of BPA may have some impacts on the developing brain at the resolution of MRI.

Chest examinations in children with real-time magnetic resonance imaging: first clinical experience

BackgroundReal-time magnetic resonance imaging (MRI) based on a fast low-angle shot technique 2.0 (FLASH 2.0) is highly effective against artifacts caused due to the bulk and pulmonary and cardiac motions of the patient. However, to date, there are no reports on the application of this innovative technique to pediatric lung MRI.ObjectiveThis study aimed to identify the limits of resolution and image quality of real-time lung MRI in children and to assess the types and minimal size of lesions with these new sequences.Materials and methodsIn this retrospective study, pathological lung findings in 87 children were classified into 6 subgroups, as detected on conventional MRI: metastases and tumors, consolidation, scars, hyperinflation, interstitial pathology and bronchiectasis. Subsequently, the findings were grouped according to size (4–6 mm, 7–9 mm and ≥ 10 mm) and evaluated for visual delineation of the findings (0 = not visible, 1 = hardly visible and 2 = well visualized).ResultsReal-time MRI allows for diagnostic, artifact-free thorax images to be obtained, regardless of patient movements. The delineation of findings strongly correlates with the size of the pathology. Metastases, consolidation and scars were visible at 100% when larger than 9 mm. In the 7–9 mm subgroup, the visibility was 83% for metastases, 88% for consolidation and 100% for scars in T2/T1 weighting. Though often visible, smaller pathological lesions of 4–6 mm in size did not regularly meet the expected diagnostic confidence: The visibility of metastases was 18%, consolidation was 64% and scars was 71%. Diffuse interstitial lung changes and hyperinflation, known as “MR-minus pathologies,” were not accessible to real-time MRI.ConclusionThe method provides motion robust images of the lung and thorax. However, the lower sensitivity for small lung lesions is a major limitation for routine use of this technique. Currently, the method is adequate for diagnosing inflammatory lung diseases, atelectasis, effusions and lung scarring in children with irregular breathing patterns or bulk motion on sedation-free MRI. A medium-term goal is to improve the diagnostic accuracy of small nodules and interstitial lesions.

Machine Learning in Prostate MRI for Prostate Cancer: Current Status and Future Opportunities.

Advances in our understanding of the role of magnetic resonance imaging (MRI) for the detection of prostate cancer have enabled its integration into clinical routines in the past two decades. The Prostate Imaging Reporting and Data System (PI-RADS) is an established imaging-based scoring system that scores the probability of clinically significant prostate cancer on MRI to guide management. Image fusion technology allows one to combine the superior soft tissue contrast resolution of MRI, with real-time anatomical depiction using ultrasound or computed tomography. This allows the accurate mapping of prostate cancer for targeted biopsy and treatment. Machine learning provides vast opportunities for automated organ and lesion depiction that could increase the reproducibility of PI-RADS categorisation, and improve co-registration across imaging modalities to enhance diagnostic and treatment methods that can then be individualised based on clinical risk of malignancy. In this article, we provide a comprehensive and contemporary review of advancements, and share insights into new opportunities in this field.

1.5 Tesla Magnetic Resonance Imaging Features of Canine Intracranial Intra-axial Hematomas.

The differentiation of solitary intra-axial hematomas from hemorrhagic neoplasms based on their magnetic resonance imaging (MRI) features is challenging. The treatment and prognosis for these two disease entities are vastly different and distinction between them is often based on MRI findings alone. The aim of this study was to describe the 1.5 tesla MRI features of canine intra-axial hematomas and correlate these findings with the evolution of hemorrhages described in human brains. Retrospective evaluation of patient details, clinical signs, and MRI findings of dogs with intra-axial hematomas that were histopathologically confirmed or determined via repeat MRI study and/or resolution of neurological signs. Ten dogs met the inclusion criteria. All 10 hematoma lesions were determined to be 2–7 days in age. On MRI, all 10 hemorrhagic lesions were comprised of two distinct regions; a relatively thin T1-weighted (T1W), T2-weighted (T2W) and gradient echo (GRE) hypointense (9/10) peripheral border region and a large central region that was heterogenous but predominantly T1W, T2W and GRE hyperintense (8/10). The peripheral border region was complete in its integrity in all 10 cases on T2W and GRE sequences. Contrast enhancement was present in (6/10) hematoma lesions and was always peripheral in nature with no evidence of central enhancement associated with any of the lesions. An intra-axial hematoma should be suspected in solitary hemorrhagic space occupying lesions that have a complete hypointense peripheral rim, elicit a peripheral contrast enhancement pattern, and display the expected temporal pattern of hematoma evolution.

New avenues for functional neuroimaging: ultra-high field MRI and OPM-MEG.

Functional brain imaging technology has developed rapidly in recent years. On the one hand, high-field 7-Tesla magnetic resonance imaging (MRI) has excelled the limited spatial resolution of 3-Tesla MRI, allowing us to enter a new world of mesoscopic imaging from the macroscopic imaging of human brain functions. On the other hand, novel optical pumping magnetometer-magnetoencephalography (OPM-MEG) has broken down the technical barriers of traditional superconducting MEG, which brings imaging of neuronal electromagnetic signals from cortical imaging to whole-brain imaging. This article aims to present a brief introduction regarding the development of conventional MRI and MEG technology, and, more importantly, to delineate that high-field MRI and OPM-MEG complement each other and together will lead us into a new era of functional brain imaging.

Contrast–enhanced ultrasound imaging of the optic nerve sheath diameter – a proof of concept study

Purpose: The present proof of concept study was performed to evaluate the consistency between measurements of the optic nerve sheath taken using contrast– enhanced sonography and high– resolution magnetic resonance imaging (MRI). The main goal was to devise a novel candidate method to measure the optic nerve sheath diameter that is easy to perform, straightforward to interpret, and highly reproducible.&#x0D; Methods: The approval of the National Medical Ethics Committee of the Republic of Slovenia was obtained; the study was registered under No. 30/03/12. Nine healthy young adults were examined with high–resolution MRI and contrast–enhanced ultrasound (CEUS). Measurements of the optic nerve sheath diameter were performed 3 mm behind the bulb, using both methods. The datasets were anonymized and the readers, blinded. Statistical analysis included evaluation of agreement using Bland–Altman plots. The assessment of inter– rater reliability was achieved by calculation of the intra class correlation coefficient.&#x0D; Results: Bland–Altman plots showed favorable overall agreement between both methods, with clinically acceptable limits of agreement. The intra class correlation coefficient calculated for both methods suggests that the CEUS method was at least as reproducible as high–resolution MRI.&#x0D; Conclusion: The described method, using CEUS to measure the optic nerve sheath diameter, facilitates the identification of the surrounding anatomy. When CEUS is employed, the measurement points can be easily and reliably set, and retro bulbar artifacts are less confounding. Based on results from a relatively small sample, this method seems to be comparable with high–resolution MRI. Evaluation of the method on a larger sample of healthy subjects and patients with increased intracranial pressure is needed.

Adjacent slices feature transformer network for single anisotropic 3D brain MRI image super-resolution

Magnetic resonance imaging (MRI) is widely used in clinical applications. However, due to the limitations in signal-to-noise ratio, physical properties of the scanner and scanning time, MRI images are usually acquired in low resolution, which restrains the accuracy of segmentation and recognition tasks. Recently, convolutional neural network (CNN) super-resolution methods have shown great potential in improving the resolution of MRI. Unfortunately, current methods neglect the data continuity and prior information of MRI images. In this paper, we handle the anisotropic 3D brain MRI images SR task as the problem of inserting new slices between adjacent in-plane slices. Then, we propose a novel adjacent slices feature transformer (ASFT) network to utilize the similarity of adjacent slices. Specifically, the backbone of the ASFT network consists of a series of stacked multi-branch features transformation and extraction (MFTE) blocks. In each MFTE block, we construct new spatial attention to focus on features from specific areas in reference branches and use channel attention to enhance the most valuable information. In addition, we propose a hybrid loss function with content and gradient information to refine the pixels and structures of the reconstructed image. We also introduce global residual learning to reduce the difficulty of network training. Experimental results show that the ASFT network gains the PSNR of 43.68 dB, 40.96 dB, and 41.22 dB with the scale factor of ×2 on the public Kirby21, ANVIL-adult, and MSSEG datasets, respectively. When compared with state-of-the-art MRI SR methods, the ASFT network achieves superior quantitative and qualitative performance.

Optimization of 4D flow MRI velocity field in the aorta with divergence-free smoothing.

Divergence-free smoothing with wall treatment (DFSwt) method is proposed for processing with four-dimensional (4D) flow magnetic resonance imaging (MRI) data of blood flows to enhance the quality of flow field with physical constraints. The new method satisfies the no-slip wall boundary condition and applies wall function of velocity profile for better estimating the velocity gradient in the near-wall region, and consequently improved wall shear stress (WSS) calculation against the issue of coarse resolution of 4D flow MRI. In the first testing case, blood flow field obtained from 4D flow MRI is well smoothed by DFSwt method. A great consistency is observed between the post-processed 4D flow MRI data and the computational fluid dynamics (CFD) data in the interested velocity field. WSS has an apparent improvement due to the proposed near-wall treatment with special wall function comparing to the result from original 4D flow MRI data or the DFS-processed data with no wall function. The other five cases also show the same performance that smoothed velocity field and improved WSS estimation are achieved on 4D flow MRI data optimized by DFSwt. The improvements will benefit the study of hemodynamics regarding the determination of location or the potential possibility of lesions.

Assessment of turbulent blood flow and wall shear stress in aortic coarctation using image-based simulations

In this study, we analyzed turbulent flows through a phantom (a 180^{circ } bend with narrowing) at peak systole and a patient-specific coarctation of the aorta (CoA), with a pulsating flow, using magnetic resonance imaging (MRI) and computational fluid dynamics (CFD). For MRI, a 4D-flow MRI is performed using a 3T scanner. For CFD, the standard k-epsilon , shear stress transport k-omega , and Reynolds stress (RSM) models are applied. A good agreement between measured and simulated velocity is obtained for the phantom, especially for CFD with RSM. The wall shear stress (WSS) shows significant differences between CFD and MRI in absolute values, due to the limited near-wall resolution of MRI. However, normalized WSS shows qualitatively very similar distributions of the local values between MRI and CFD. Finally, a direct comparison between in vivo 4D-flow MRI and CFD with the RSM turbulence model is performed in the CoA. MRI can properly identify regions with locally elevated or suppressed WSS. If the exact values of the WSS are necessary, CFD is the preferred method. For future applications, we recommend the use of the combined MRI/CFD method for analysis and evaluation of the local flow patterns and WSS in the aorta.

Molecular diagnostics in drug-resistant focal epilepsy define new disease entities.

Structural brain lesions, including the broad range of malformations of cortical development (MCD) and glioneuronal tumors, are among the most common causes of drug‐resistant focal epilepsy. Epilepsy surgery can provide a curative treatment option in respective patients. The currently available pre‐surgical multi‐modal diagnostic armamentarium includes high‐ and ultra‐high resolution magnetic resonance imaging (MRI) and intracerebral EEG to identify a focal structural brain lesion as epilepsy underlying etiology. However, specificity and accuracy in diagnosing the type of lesion have proven to be limited. Moreover, the diagnostic process does not stop with the decision for surgery. The neuropathological diagnosis remains the gold standard for disease classification and patient stratification, but is particularly complex with high inter‐observer variability. Here, the identification of lesion‐specific mosaic variants together with epigenetic profiling of lesional brain tissue became new tools to more reliably identify disease entities. In this review, we will discuss how the paradigm shifts from histopathology toward an integrated diagnostic approach in cancer and the more recent development of the DNA methylation‐based brain tumor classifier have started to influence epilepsy diagnostics. Some examples will be highlighted showing how the diagnosis and our mechanistic understanding of difficult to classify structural brain lesions associated with focal epilepsy has improved with molecular genetic data being considered in decision making.

Super Resolution of Magnetic Resonance Images.

In this work, novel denoising and super resolution (SR) approaches for magnetic resonance (MR) images are addressed, and are integrated in a unified framework, which do not require example low resolution (LR)/high resolution (HR)/cross-modality/noise-free images and prior information of noise–noise variance. The proposed method categorizes the patches as either smooth or textured and then denoises them by deploying different denoising strategies for efficient denoising. The denoising algorithm is integrated into the SR approach, which uses a gradient profile-based constraint in a sparse representation-based framework to improve the resolution of MR images with reduced smearing of image details. This constraint regularizes the estimation of HR images such that the estimated HR image has gradient profiles similar to the gradient profiles of the original HR image. For this, the gradient profile sharpness (GPS) values of an unknown HR image are estimated using an approximated piece-wise linear relation among GPS values of LR and upsampled LR images. The experiments are performed on three different publicly available datasets. The proposed SR approach outperforms the existing unsupervised SR approach addressed for real MR images that exploits low rank and total variation (LRTV) regularization, by an average peak signal to noise ratio (PSNR) of 0.73 dB and 0.38 dB for upsampling factors 2 and 3, respectively. For the super resolution of noisy real MR images (degraded with 2% noise), the proposed approach outperforms the LRTV approach by an average PSNR of 0.54 dB and 0.46 dB for upsampling factors 2 and 3, respectively. The qualitative analysis is shown for real MR images from healthy subjects and subjects with Alzheimer’s disease and structural deformity, i.e., cavernoma.

3D MRI Reconstruction Based on 2D Generative Adversarial Network Super-Resolution.

The diagnosis of brain pathologies usually involves imaging to analyze the condition of the brain. Magnetic resonance imaging (MRI) technology is widely used in brain disorder diagnosis. The image quality of MRI depends on the magnetostatic field strength and scanning time. Scanners with lower field strengths have the disadvantages of a low resolution and high imaging cost, and scanning takes a long time. The traditional super-resolution reconstruction method based on MRI generally states an optimization problem in terms of prior information. It solves the problem using an iterative approach with a large time cost. Many methods based on deep learning have emerged to replace traditional methods. MRI super-resolution technology based on deep learning can effectively improve MRI resolution through a three-dimensional convolutional neural network; however, the training costs are relatively high. In this paper, we propose the use of two-dimensional super-resolution technology for the super-resolution reconstruction of MRI images. In the first reconstruction, we choose a scale factor of 2 and simulate half the volume of MRI slices as input. We utilize a receiving field block enhanced super-resolution generative adversarial network (RFB-ESRGAN), which is superior to other super-resolution technologies in terms of texture and frequency information. We then rebuild the super-resolution reconstructed slices in the MRI. In the second reconstruction, the image after the first reconstruction is composed of only half of the slices, and there are still missing values. In our previous work, we adopted the traditional interpolation method, and there was still a gap in the visual effect of the reconstructed images. Therefore, we propose a noise-based super-resolution network (nESRGAN). The noise addition to the network can provide additional texture restoration possibilities. We use nESRGAN to further restore MRI resolution and high-frequency information. Finally, we achieve the 3D reconstruction of brain MRI images through two super-resolution reconstructions. Our proposed method is superior to 3D super-resolution technology based on deep learning in terms of perception range and image quality evaluation standards.

Brain MRI super-resolution using coupled-projection residual network

Magnetic Resonance Imaging (MRI) has been widely used in clinical application and pathology research to help doctors provide better diagnoses. However, accurate diagnosis by MRI remains a great challenge, as images obtained via current MRI techniques usually have low resolutions. Improving MRI image quality and resolution has thus become a critically important task. This paper presents an innovative Coupled-Projection Residual Network (CPRN) for MRI super-resolution. CPRN consists of two complementary sub-networks: a shallow network and a deep one, which maintain content consistency while learning high frequency differences between low-resolution and high-resolution images. The shallow sub-network employs coupled-projection to better retain the MR image details, where a novel feedback mechanism is introduced to guide the reconstruction of high-resolution images. The deep sub-network learns from the residuals of the high-frequency image information, where multiple residual blocks are cascaded to magnify the MR images at the last network layer. Finally, the features from the shallow and deep sub-networks are fused for the reconstruction of high-resolution MR images. For effective feature fusion between the deep and shallow sub-networks, a step-wise connection (CPRN_S) is designed, inspired by the human cognitive process (from simple to complex). Experiments over three public MRI datasets show that our proposed CPRN achieves superior MRI super-resolution performance compared with the state-of-the-art.