Real Magnetic Resonance Imaging Research Articles

Efficient dual ADMMs for sparse compressive sensing MRI reconstruction

Magnetic Resonance Imaging (MRI) is a kind of medical imaging technology used for diagnostic imaging of diseases, but its image quality may be suffered by the long acquisition time. The compressive sensing (CS) based strategy may decrease the reconstruction time greatly, but it needs efficient reconstruction algorithms to produce high-quality and reliable images. This paper focuses on the algorithmic improvement for the sparse reconstruction of CS-MRI, especially considering a non-smooth convex minimization problem which is composed of the sum of a total variation regularization term and a $$\ell _1$$ -norm term of the wavelet transformation. The partly motivation of targeting the dual problem is that the dual variables are involved in relatively low-dimensional subspace. Instead of solving the primal model as usual, we turn our attention to its associated dual model composed of three variable blocks and two separable non-smooth function blocks. However, the directly extended alternating direction method of multipliers (ADMM) must be avoided because it may be divergent, although it usually performs well numerically. In order to solve the problem, we employ a symmetric Gauss–Seidel (sGS) technique based ADMM. Compared with the directly extended ADMM, this method only needs one additional iteration, but its convergence can be guaranteed theoretically. Besides, we also propose a generalized variant of ADMM because this method has been illustrated to be efficient for solving semidefinite programming in the past few years. Finally, we do extensive experiments on MRI reconstruction using some simulated and real MRI images under different sampling patterns and ratios. The numerical results demonstrate that the proposed algorithms significantly achieve high reconstruction accuracies with fast computational speed.

Effect of field strength on RF power deposition near conductive leads: A simulation study of SAR in DBS lead models during MRI at 1.5 T-10.5 T.

Since the advent of magnetic resonance imaging (MRI) nearly four decades ago, there has been a quest for ever-higher magnetic field strengths. Strong incentives exist to do so, as increasing the magnetic field strength increases the signal-to-noise ratio of images. However, ensuring patient safety becomes more challenging at high and ultrahigh field MRI (i.e., ≥3 T) compared to lower fields. The problem is exacerbated for patients with conductive implants, such as those with deep brain stimulation (DBS) devices, as excessive local heating can occur around implanted lead tips. Despite extensive effort to assess radio frequency (RF) heating of implants during MRI at 1.5 T, a comparative study that systematically examines the effects of field strength and various exposure limits on RF heating is missing. This study aims to perform numerical simulations that systematically compare RF power deposition near DBS lead models during MRI at common clinical and ultra-high field strengths, namely 1.5, 3, 7, and 10.5 T. Furthermore, we assess the effects of different exposure constraints on RF power deposition by imposing limits on either the B1+ or global head specific absorption rate (SAR) as these two exposure limits commonly appear in MRI guidelines. We created 33 unique DBS lead models based on postoperative computed tomography (CT) images of patients with implanted DBS devices and performed electromagnetic simulations to evaluate the SAR of RF energy in the tissue surrounding lead tips during RF exposure at frequencies ranging from 64 MHz (1.5 T) to 447 MHz (10.5 T). The RF exposure was implemented via realistic MRI RF coil models created based on physical prototypes built in our institutions. We systematically examined the distribution of local SAR at different frequencies with the input coil power adjusted to either limit the B1+ or the global head SAR. The MRI RF coils at higher resonant frequencies generated lower SARs around the lead tips when the global head SAR was constrained. The trend was reversed when the constraint was imposed on B1+. At higher static fields, MRI is not necessarily more dangerous than at lower fields for patients with conductive leads. Specifically, when a conservative safety criterion, such as constraints on the global SAR, is imposed, coils at a higher resonant frequency tend to generate a lower local SAR around implanted leads due to the decreased B1+ and, by proxy, E field levels.

Deformable cardiac surface tracking by adaptive estimation algorithms

This study presents a particle filter based framework to track cardiac surface from a time sequence of single magnetic resonance imaging (MRI) slices with the future goal of utilizing the presented framework for interventional cardiovascular magnetic resonance procedures, which rely on the accurate and online tracking of the cardiac surface from MRI data. The framework exploits a low-order parametric deformable model of the cardiac surface. A stochastic dynamic system represents the cardiac surface motion. Deformable models are employed to introduce shape prior to control the degree of the deformations. Adaptive filters are used to model complex cardiac motion in the dynamic model of the system. Particle filters are utilized to recursively estimate the current state of the system over time. The proposed method is applied to recover biventricular deformations and validated with a numerical phantom and multiple real cardiac MRI datasets. The algorithm is evaluated with multiple experiments using fixed and varying image slice planes at each time step. For the real cardiac MRI datasets, the average root-mean-square tracking errors of 2.61 mm and 3.42 mm are reported respectively for the fixed and varying image slice planes. This work serves as a proof-of-concept study for modeling and tracking the cardiac surface deformations via a low-order probabilistic model with the future goal of utilizing this method for the targeted interventional cardiac procedures under MR image guidance. For the real cardiac MRI datasets, the presented method was able to track the points-of-interests located on different sections of the cardiac surface within a precision of 3 pixels. The analyses show that the use of deformable cardiac surface tracking algorithm can pave the way for performing precise targeted intracardiac ablation procedures under MRI guidance. The main contributions of this work are twofold. First, it presents a framework for the tracking of whole cardiac surface from a time sequence of single image slices. Second, it employs adaptive filters to incorporate motion information in the tracking of nonrigid cardiac surface motion for temporal coherence.

A Hybrid Volume-Surface Integral Equation Method for Rapid Electromagnetic Simulations in MRI.

We developed a hybrid volume surface integral equation (VSIE) method based on domain decomposition to perform fast and accurate magnetic resonance imaging (MRI) simulations that include both remote and local conductive elements. We separated the conductive surfaces present in MRI setups into two domains and optimized electromagnetic (EM) modeling for each case. Specifically, interactions between the body and EM waves originating from local radiofrequency (RF) coils were modeled with the precorrected fast Fourier transform, whereas the interactions with remote conductive surfaces (RF shield, scanner bore) were modeled with a novel cross tensor train-based algorithm. We compared the hybrid-VSIE with other VSIE methods for realistic MRI simulation setups. The hybrid-VSIE was the only practical method for simulation using 1 mm voxel isotropic resolution (VIR). For 2 mm VIR, our method could be solved at least 23 times faster and required 760 times lower memory than traditional VSIE methods. The hybrid-VSIE demonstrated a marked improvement in terms of convergence times of the numerical EM simulation compared to traditional approaches in multiple realistic MRI scenarios. The efficiency of the novel hybrid-VSIE method could enable rapid simulations of complex and comprehensive MRI setups.

MRI Generated from CT for Acute Ischemic Stroke Combining Radiomics and Generative Adversarial Networks.

Compared to computed tomography (CT), magnetic resonance imaging (MRI) is more sensitive to acute ischemic stroke lesion. However, MRI is time-consuming, expensive, and susceptible to interference from metal implants. Generating MRI images from CT images can address the limitations of MRI. The key problem in the process is obtaining lesion information from CT. In this study, we propose a cross-modal image generation algorithm from CT to MRI for acute ischemic stroke by combining radiomics with generative adversarial networks. First, the lesion candidate region was obtained using radiomics, the radiomic features of the region were extracted, and the feature with the largest information gain was selected and visualized as a feature map. Then, the concatenation of the extracted feature map and the CT image was input in the generator. We added a residual module after the downsampling of the generator, following the general shape of U-Net, which can deepen the network without causing degradation problems. In addition, we introduced the lesion feature similarity loss function to focus the model on the similarity of the lesion. Through the subjective judgment of two experienced radiologists and using evaluation metrics, the results showed that the generated MRI images were very similar to the real MRI images. Moreover, the locations of the lesions were correct, and the shapes of lesions were similar to those of the real lesions, which can help doctors with timely diagnosis and treatment.

Real-time image-guided treatment of mobile tumors in proton therapy by a library of treatment plans: a simulation study.

To improve target coverage and reduce the dose in the surrounding organs-at-risks (OARs), we developed an image-guided treatment method based on a precomputed library of treatment plans controlled and delivered in real-time. A library of treatment plans is constructed by optimizing a plan for each breathing phase of a four dimensional computed tomography (4DCT). Treatments are delivered by simulation on a continuous sequence of synthetic computed tomographies (CTs) generated from real magnetic resonance imaging (MRI) sequences. During treatment, the plans for which the tumor are at a close distance to the current tumor position are selected to deliver their spots. The study is conducted on five livercases. We tested our approach under imperfect knowledge of the tumor positions with a 2 mm distance error. On average, compared to a 4D robustly optimized treatment plan, our approach led to a dose homogeneity increase of 5% (defined as ) in the target and a mean liver dose decrease of 23%. The treatment time was roughly increased by a factor of 2 but remained below 4 min onaverage. Our image-guided treatment framework outperforms state-of-the-art 4D-robust plans for all patients in this study on both target coverage and OARs sparing, with an acceptable increase in treatment time under the current accuracy of the tumor trackingtechnology.

Modeling of glioma growth using modified reaction-diffusion equation on brain MR images

Background and Objective: Modeling of glioma growth and evolution is of key importance for cancer diagnosis, predicting clinical progression and improving treatment outcomes of neurosurgery. However, existing models are unable to characterize spatial variations of the proliferation and infiltration of tumor cells, making it difficult to achieve accurate prediction of tumor growth. Methods: In this paper, a new growth model of brain tumor using a reaction-diffusion equation on brain magnetic resonance images is proposed. Both the heterogeneity of brain tissue and the density of tumor cells are used to estimate the proliferation and diffusion coefficients of brain tumor cells. The diffusion coefficient that characterizes tumor diffusion and infiltration is calculated based on the ratio of tissues (white and gray matter), while the proliferation coefficient is evaluated using the spatial gradient of tumor cells. In addition, a parameter space is constructed using inverse distance weighted interpolation to describe the spatial distribution of proliferation coefficient.Results: The glioma growth predicted by the proposed model were tested by comparing with the real magnetic resonance images of the patients. Experiments and simulation results show that the proposed method achieves accurate modeling of glioma growth. The interpolation-based growth model has higher average dice score of 0.0647 and 0.0545, and higher average Jaccard index of 0.0673 and 0.0573, respectively, compared to the uniform- and gradient-based growth models. Conclusions: The experimental results demonstrate the feasibility of calculating the proliferation and diffusion coefficients of the growth model based on patient-specific anatomy. The parameter space that characterizes spatial distribution of proliferation and diffusion coefficients is established and incorporated into the simulation of glioma growth. It enables to obtain patient-specific models about glioma growth by estimating and calibrating only a few model parameters.

PTNet3D: A 3D High-Resolution Longitudinal Infant Brain MRI Synthesizer Based on Transformers.

An increased interest in longitudinal neurodevelopment during the first few years after birth has emerged in recent years. Noninvasive magnetic resonance imaging (MRI) can provide crucial information about the development of brain structures in the early months of life. Despite the success of MRI collections and analysis for adults, it remains a challenge for researchers to collect high-quality multimodal MRIs from developing infant brains because of their irregular sleep pattern, limited attention, inability to follow instructions to stay still during scanning. In addition, there are limited analytic approaches available. These challenges often lead to a significant reduction of usable MRI scans and pose a problem for modeling neurodevelopmental trajectories. Researchers have explored solving this problem by synthesizing realistic MRIs to replace corrupted ones. Among synthesis methods, the convolutional neural network-based (CNN-based) generative adversarial networks (GANs) have demonstrated promising performance. In this study, we introduced a novel 3D MRI synthesis framework- pyramid transformer network (PTNet3D)- which relies on attention mechanisms through transformer and performer layers. We conducted extensive experiments on high-resolution Developing Human Connectome Project (dHCP) and longitudinal Baby Connectome Project (BCP) datasets. Compared with CNN-based GANs, PTNet3D consistently shows superior synthesis accuracy and superior generalization on two independent, large-scale infant brain MRI datasets. Notably, we demonstrate that PTNet3D synthesized more realistic scans than CNN-based models when the input is from multi-age subjects. Potential applications of PTNet3D include synthesizing corrupted or missing images. By replacing corrupted scans with synthesized ones, we observed significant improvement in infant whole brain segmentation.

CSR-PERT: Joint framework for MRI and HARDI data reconstruction using perturbed radial trajectory estimated from compressively sensed measurements

Radial sampling pattern is an important signal acquisition strategy in magnetic resonance imaging (MRI) owing to better immunity to motion-induced artifacts and less pronounced aliasing due to undersampling compared to the Cartesian sampling. These advantages of radial sampling can be exploited in acquisition of multidimensional signals such as High Angular Resolution Diffusion Imaging (HARDI), with tremendous scope of acceleration. Despite such benefits, gradient delays lead to samples being acquired from unknown miscentered radial trajectories, severely degrading the image reconstruction quality. In the present work, we propose Csr-Pert that is a joint framework, wherein these perturbed radial trajectories are estimated and utilized for image reconstruction from the compressively sensed measurements of (i) MRI data and (ii) HARDI data. The proposed Csr-Pert method is tested on one real MRI dataset with trajectory deviations and is observed to perform better than the existing state-of-the-art method at high acceleration factors up to 8. To the best of our knowledge, this is the first work to address the problem of estimating perturbed trajectories using the compressively sensed MRI and HARDI data. The method is also tested for varying combinations of trajectory deviations and sampling proportions. It is observed to yield very good quality HARDI reconstruction for a wide variety of scenarios. We have also demonstrated the robustness of the proposed method on real datasets in clinical settings assuming perturbed as well as noisy trajectories.

Multimodality-based super-resolution reconstruction for routine brain magnetic resonance images

To propose a multi-modality-based super-resolution synthesis model for reconstruction of routine brain magnetic resonance images (MRI) with a low resolution and a high thickness into high-resolution images. Based on real paired low-high resolution MRI data (2D T1, 2D T2 FLAIR and 3D T1), a structure-constrained image mapping network was used to extract important features from the images with different modalities including the whole T1 and subcortical regions of T2 FLAIR to reconstruct T1 images with higher resolutions. The gray scale intensity and structural similarities between the super-resolution images and high-resolution images were used to enhance the reconstruction performance. We used the anatomical information acquired from segment maps of the super-resolution T1 image and the ground truth by a segmentation tool as a significant constraint for adaptive learning of the intrinsic tissue structure characteristics of the brain to improve the reconstruction performance of the model. Our method showed the performance on the testing dataset than other methods with an average PSNR of 33.11 and SSIM of 0.996. The anatomical structure of the brain including the sulcus, gyrus, and subcortex were all reconstructed clearly using the proposed method, which also greatly enhanced the precision of MSCSR for brain volume measurement. The proposed MSCSR model shows excellent performance for reconstructing super-resolution brain MR images based on the information of brain tissue structure and multimodality MR images.

EEG functional connectivity analysis in the source space

There is a growing interest in using electroencephalography (EEG) and source modeling to investigate functional interactions among cortical processes, particularly when dealing with pediatric populations. This paper introduces two pipelines that have been recently used to conduct EEG FC analysis in the cortical source space. The analytic streams of these pipelines can be summarized into the following steps: 1) cortical source reconstruction of high-density EEG data using realistic magnetic resonance imaging (MRI) models created with age-appropriate MRI templates; 2) segmentation of reconstructed source activities into brain regions of interest; and 3) estimation of FC in age-related frequency bands using robust EEG FC measures, such as weighted phase lag index and orthogonalized power envelope correlation. In this paper we demonstrate the two pipelines with resting-state EEG data collected from children at 12 and 36 months of age. We also discuss the advantages and limitations of the methods/techniques integrated into the pipelines. Given there is a need in the research community for open-access analytic toolkits that can be used for pediatric EEG data, programs and codes used for the current analysis are made available to the public.

Generation of Synthetic Rat Brain MRI Scans with a 3D Enhanced Alpha Generative Adversarial Network

Translational brain research using Magnetic Resonance Imaging (MRI) is becoming increasingly popular as animal models are an essential part of scientific studies and more ultra-high-field scanners are becoming available. Some disadvantages of MRI are the availability of MRI scanners and the time required for a full scanning session. Privacy laws and the 3Rs ethics rule also make it difficult to create large datasets for training deep learning models. To overcome these challenges, an adaptation of the alpha Generative Adversarial Networks (GANs) architecture was used to test its ability to generate realistic 3D MRI scans of the rat brain in silico. As far as the authors are aware, this was the first time a GAN-based approach was used to generate synthetic MRI data of the rat brain. The generated scans were evaluated using various quantitative metrics, a Turing test, and a segmentation test. The last two tests proved the realism and applicability of the generated scans to real problems. Therefore, by using the proposed new normalisation layer and loss functions, it was possible to improve the realism of the generated rat MRI scans, and it was shown that using the generated data improved the segmentation model more than using the conventional data augmentation.

Quality assessment of anatomical MRI images from generative adversarial networks: Human assessment and image quality metrics

BackgroundGenerative Adversarial Networks (GANs) can synthesize brain images from image or noise input. So far, the gold standard for assessing the quality of the generated images has been human expert ratings. However, due to limitations of human assessment in terms of cost, scalability, and the limited sensitivity of the human eye to more subtle statistical relationships, a more automated approach towards evaluating GANs is required. New methodWe investigated to what extent visual quality can be assessed using image quality metrics and we used group analysis and spatial independent components analysis to verify that the GAN reproduces multivariate statistical relationships found in real data. Reference human data was obtained by recruiting neuroimaging experts to assess real Magnetic Resonance (MR) images and images generated by a GAN. Image quality was manipulated by exporting images at different stages of GAN training. ResultsExperts were sensitive to changes in image quality as evidenced by ratings and reaction times, and the generated images reproduced group effects (age, gender) and spatial correlations moderately well. We also surveyed a number of image quality metrics. Overall, Fréchet Inception Distance (FID), Maximum Mean Discrepancy (MMD) and Naturalness Image Quality Evaluator (NIQE) showed sensitivity to image quality and good correspondence with the human data, especially for lower-quality images (i.e., images from early stages of GAN training). However, only a Deep Quality Assessment (QA) model trained on human ratings was able to reproduce the subtle differences between higher-quality images. ConclusionsWe recommend a combination of group analyses, spatial correlation analyses, and both distortion metrics (FID, MMD, NIQE) and perceptual models (Deep QA) for a comprehensive evaluation and comparison of brain images produced by GANs.

Denoising 3D magnetic resonance images based on weighted tensor nuclear norm minimization using balanced nonlocal patch tensors

The denoising of magnetic resonance (MR) images is important to improve the accuracy of organ tissue information recognition in the process of medical diagnosis. This paper proposes a 3D MR image denoising method based on weighted tensor nuclear norm minimization using balanced nonlocal patch tensors. In order to make utilization of the nonlocal self-similarity in high-dimensional images and the correlation among different dimensions, a high-dimensional adaptive clustering technology is developed to construct highly correlated 3D MR image patches into nonlocal patch tensors. Since the generated tensor exhibits low-rank characteristics, a low-rank tensor approximation technique based on the tensor singular value decomposition framework can be used to denoise the MR data. To improve the denoising accuracy, the local noise level information of different nonlocal patch tensors is considered. Besides, a method for constructing more balanced 3D nonlocal patch tensors is proposed to exploit the effectiveness of the tensor singular value decomposition. Experimental results show that the proposed method has a greater improvement over the existing MR image denoising methods in terms of objective measurement and subjective visual quality on simulated and real 3D MR images with different categories and noise levels.

Dosimetric Effects of Air Cavities for MRI-Guided Online Adaptive Radiation Therapy (MRgART) of Prostate Bed after Radical Prostatectomy.

Purpose: To evaluate dosimetric impact of air cavities and their corresponding electron density correction for 0.35 tesla (T) Magnetic Resonance-guided Online Adaptive Radiation Therapy (MRgART) of prostate bed patients. Methods: Three 0.35 T MRgRT plans (anterior–posterior (AP) beam, AP–PA beams, and clinical intensity modulated radiation therapy (IMRT)) were generated on a prostate bed patient’s (Patient A) planning computed tomography (CT) with artificial rectal air cavities of various sizes (0–3 cm, 0.5 cm increments). Furthermore, two 0.35 T MRgART plans (‘Deformed’ and ‘Override’) were generated on a prostate bed patient’s (Patient B) daily magnetic resonance image (MRI) with artificial rectal air cavities of various sizes (0–3 cm, 0.5 cm increments) and on five prostate bed patient’s (Patient 1–5) daily MRIs (2 MRIs: Fraction A and B) with real air cavities. For each MRgART plan, daily MRI electron density map was obtained by deformable registration with simulation CT. In the ‘Deformed’ plan, a clinical IMRT plan is calculated on the daily MRI with electron density map obtained from deformable registration only. In the ‘Override’ plan, daily MRI and simulation CT air cavities are manually corrected and bulk assigned air and water density on the registered electron density map, respectively. Afterwards, the clinical IMRT plan is calculated. Results: For the MRgRT plans, AP and AP–PA plans’ rectum/rectal wall max dose increased with increasing air cavity size, where the 3 cm air cavity resulted in a 20%/17% and 13%/13% increase, relative to no air cavity, respectively. Clinical IMRT plan was robust to air cavity size, where dose change remained less than 1%. For the MRgART plans, daily MRI electron density maps, obtained from deformable registration with simulation CT, was unable to accurately produce electron densities reflecting the air cavities. However, for the artificial daily MRI air cavities, dosimetric change between ‘Deformed’ and ‘Override’ plan was small (<4%). Similarly, for the real daily MRI air cavities, clinical constraint changes between ‘Deformed’ and ‘Override’ plan was negligible and did not lead to change in clinical decision for adaptive planning except for two fractions. In these fractions, the ‘Override’ plan indicated that the bladder max dose and rectum V35.7 exceeded the constraint, while the ‘Deformed’ plan showed acceptable dose, although the absolute difference was only 0.3 Gy and 0.03 cc, respectively. Conclusion: Clinical 0.35 T IMRT prostate bed plans are dosimetrically robust to air cavities. MRgART air cavity electron density correction shows clinically insignificant change and is not warranted on low-field systems.

Epilepsy Radiology Reports Classification Using Deep Learning Networks

The automatic and accurate classification of Magnetic Resonance Imaging (MRI) radiology report is essential for the analysis and interpretation epilepsy and non-epilepsy. Since the majority of MRI radiology reports are unstructured, the manual information extraction is time-consuming and requires specific expertise. In this paper, a comprehensive method is proposed to classify epilepsy and non-epilepsy real brain MRI radiology text reports automatically. This method combines the Natural Language Processing technique and statistical Machine Learning methods. 122 real MRI radiology text reports (97 epilepsy, 25 non-epilepsy) are studied by our proposed method which consists of the following steps: (i) for a given text report our systems first cleans HTML/XML tags, tokenize, erase punctuation, normalize text, (ii) then it converts into MRI text reports numeric sequences by using index-based word encoding, (iii) then we applied the deep learning models that are uni-directional long short-term memory (LSTM) network, bidirectional long short-term memory (BiLSTM) network and convolutional neural network (CNN) for the classifying comparison of the data, (iv) finally, we used 70% of used for training, 15% for validation, and 15% for test observations. Unlike previous methods, this study encompasses the following objectives: (a) to extract significant text features from radiologic reports of epilepsy disease; (b) to ensure successful classifying accuracy performance to enhance epilepsy data attributes. Therefore, our study is a comprehensive comparative study with the epilepsy dataset obtained from numeric sequences by using index-based word encoding method applied for the deep learning models. The traditional method is numeric sequences by using index-based word encoding which has been made for the first time in the literature, is successful feature descriptor in the epilepsy data set. The BiLSTM network has shown a promising performance regarding the accuracy rates. We show that the larger sized medical text reports can be analyzed by our proposed method.

Multi-modal brain MRI images enhancement based on framelet and local weights super-resolution.

Magnetic resonance (MR) image enhancement technology can reconstruct high-resolution image from a low-resolution image, which is of great significance for clinical application and scientific research. T1 weighting and T2 weighting are the two common magnetic resonance imaging modes, each of which has its own advantages, but the imaging time of T2 is much longer than that of T1. Related studies have shown that they have very similar anatomical structures in brain images, which can be utilized to enhance the resolution of low-resolution T2 images by using the edge information of high-resolution T1 images that can be rapidly imaged, so as to shorten the imaging time needed for T2 images. In order to overcome the inflexibility of traditional methods using fixed weights for interpolation and the inaccuracy of using gradient threshold to determine edge regions, we propose a new model based on previous studies on multi-contrast MR image enhancement. Our model uses framelet decomposition to finely separate the edge structure of the T2 brain image, and uses the local regression weights calculated from T1 image to construct a global interpolation matrix, so that our model can not only guide the edge reconstruction more accurately where the weights are shared, but also carry out collaborative global optimization for the remaining pixels and their interpolated weights. Experimental results on a set of simulated MR data and two sets of real MR images show that the enhanced images obtained by the proposed method are superior to the compared methods in terms of visual sharpness or qualitative indicators.

Super-resolution method for MR images based on multi-resolution CNN

High-Resolution (HR) Magnetic Resonance Images (MRI) can help physician diagnosis lesion more effectively. However, in practice, it is difficult to obtain HR-MRI due to equipment limitations, scanning time or patient comfort. Fortunately, with the development of information technology, HR-MRI could be obtained by some image post-processing approaches. This paper presents a novel HR-MRI generation approach based on convolution neural network (CNN) and multi-resolution analysis, which intends to improve the resolution of the low-resolution (LR) T2-weighted (T2w) images with the prior information provided by HR T1-weighted (T1w) reference images based on the inherent correlation between the two. Specifically, a novel 3D multi-resolution analysis multi-modality SR reconstruction network for MRI is built, which takes full advantage of structural similarity between the modalities. Different from other networks, the proposed reconstruction network fully combines the idea of multi-resolution analysis. Firstly, we built an associative memory network between HR-T1w and HR-T2w, which learns high-frequency feature from HR-T1w for HR-T2w at different scales. Subsequently, a progressive reconstruction approach is presented to combine the multi-scale high-frequency feature extracted by the first step from HR-T1w with the down-sampling 4x LR-T2w feature to generate 2× and 4× HR-T2w images, respectively. The experimental results on two real MRI dataset show the effectiveness of the proposed 3D SR reconstruction network, which achieves improved reconstruction performance compared with other single- and multi-modal networks. Even with 4x magnification, our method can effectively restore the edge details of the image.

Virtual Contrast-Enhanced Magnetic Resonance Images Synthesis for Patients With Nasopharyngeal Carcinoma Using Multimodality-Guided Synergistic Neural Network

To investigate a novel deep-learning network that synthesizes virtual contrast-enhanced T1-weighted (vceT1w) magnetic resonance images (MRI) from multimodality contrast-free MRI for patients with nasopharyngeal carcinoma (NPC). This article presents a retrospective analysis of multiparametric MRI, with and without contrast enhancement by gadolinium-based contrast agents (GBCAs), obtained from 64 biopsy-proven cases of NPC treated at Hong Kong Queen Elizabeth Hospital. A multimodality-guided synergistic neural network (MMgSN-Net) was developed to leverage complementary information between contrast-free T1-weighted and T2-weighted MRI for vceT1w MRI synthesis. Thirty-five patients were randomly selected for model training, whereas 29 patients were selected for model testing. The synthetic images generated from MMgSN-Net were quantitatively evaluated against real GBCA-enhanced T1-weighted MRI using a series of statistical evaluating metrics, which include mean absolute error (MAE), mean squared error (MSE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR). Qualitative visual assessment between the real and synthetic MRI was also performed. Effectiveness of our MMgSN-Net was compared with 3 state-of-the-art deep-learning networks, including U-Net, CycleGAN, and Hi-Net, both quantitatively and qualitatively. Furthermore, a Turing test was performed by 7 board-certified radiation oncologists from 4 hospitals for assessing authenticity of the synthesized vceT1w MRI against the real GBCA-enhanced T1-weighted MRI. Results from the quantitative evaluations demonstrated that our MMgSN-Net outperformed U-Net, CycleGAN and Hi-Net, yielding the top-ranked scores in averaged MAE (44.50 ± 13.01), MSE (9193.22 ± 5405.00), SSIM (0.887 ± 0.042), and PSNR (33.17 ± 2.14). Furthermore, the mean accuracy of the 7 readers in the Turing tests was determined to be 49.43%, equivalent to random guessing (ie, 50%) in distinguishing between real GBCA-enhanced T1-weighted and synthetic vceT1w MRI. Qualitative evaluation indicated that MMgSN-Net gave the best approximation to the ground-truth images, particularly in visualization of tumor-to-muscle interface and the intratumor texture information. Our MMgSN-Net was capable of synthesizing highly realistic vceT1w MRI that outperformed the 3 comparable state-of-the-art networks.

Segmentation of Brain MR Image Using Modified Student’s t-Mixture Model

In conventional brain image analysis, it is a critical step to segment brain magnetic resonance (MR) image into three major tissues: Gray Matter (GM), White Matter (WM) and Cerebrospinal Fluid (CSF). The main difficulties for segmenting brain MR image are partial volume effect, intensity inhomogeneity and noise, which result in challenging segmentation task. In this paper, we propose a novel modified method based on the basis of the conventional Student’s t-Mixture Model (SMM), for segmentation of brain MR image and correction of bias field. The advantages of our model are introduced as follows. First, we take account of the influence on the probabilities of the pixels in the adjacent region and take full advantage of the local spatial information and class information. Second, our modified SMM is derived from the traditional finite mixture model (FMM) by adding the bias field correction model; the logarithmic likelihood function of traditional FMM is revised. Third, the noise and bias field can be easily extended to combine with the SMM model and EM algorithm. Last but not least, the exponential coefficients are employed to control the results of segmentation details. As a result, our effective and highly accurate method exhibits high robustness on both simulated and real MR image segmentation, compared to the state-of-the-art algorithms.