Magnitude Images Research Articles

Leveraging Phase Information of 3D Isotropic Ultrashort Echo Time Pulmonary MRI for the Detection of Thoracic Lymphadenopathy

Background Ultrashort echo time (UTE) allows imaging of tissues with short relaxation times, but it comes with the expense of long scan times. Magnitude images of UTE magnetic resonance imaging (MRI) are widely used in pulmonary imaging due to excellent parenchymal signal, but have insufficient contrast for other anatomical regions of the thorax. Our work investigates the value of UTE phase images (UTE-Ps)—generated simultaneously from the acquired UTE signal used for the magnitude images—for the detection of thoracic lymph nodes based on water-fat contrast. It employs an advanced imaging sequence and reconstruction allowing isotropic 3D UTE MRI in clinically acceptable scan times. Methods In our prospective study, 42 patients with 136 lymph nodes had undergone venous computed tomography and pulmonary MRI scans with UTE within a 14-day interval. 3D isotropic UTE images were acquired using FLORET (fermat looped, orthogonally encoded trajectories). Background-corrected phase images (UTE-P) and magnitude images were reconstructed simultaneously from the UTE-Signal. Three radiologists performed a blinded reading in which all lymph nodes with a short-axis diameter (SAD) of at least 0.5 cm were detected. Detection rates and performance metrics of UTE-P for all lymph node regions and for pathologic (SAD ≥10 mm) and nonpathologic lymph nodes (SAD <10 mm) were calculated using computed tomography as a reference. The interreader agreement defined as the presence or absence of lymph nodes based on patient and region was calculated using Fleiss kappa (κ). Findings In the phase images, pathologic lymph nodes in the mediastinal and hilar region were detected with a high diagnostic confidence due to the achieved water-fat contrast (average sensitivity, specificity, positive predictive value, and negative predictive value of 95.83% [confidence interval (CI), 92.76%–98.91%], 100%, 100%, and 99.32% [CI, 98.08%–100%]). Stepwise inclusion of all lymph node regions and nonpathologic lymph nodes was associated with a moderate decrease resulting in an average sensitivity, specificity, positive predictive value, and negative predictive value of 77.9% (CI, 70.9%–84.7%), 99.4% (CI, 98.7%–99.9%), 97.0% (CI, 93.4%–99.7%), and 94.7% (CI, 92.8%–96.4%) for the inclusion of all lymph nodes sizes and regions. Interreader agreement was almost perfect (κ = 0.92). Conclusions Pathological lymph nodes in the mediastinal and hilar region can be detected in phase-images with high diagnostic confidence, thanks to the ability of the phase images to depict water-fat contrast in combination with high spatial 3D resolution, extending the clinical applicability of UTE into the simultaneous assessment of lung parenchyma and lymph nodes.

MR‐zero meets FLASH – controlling the transient signal decay in gradient‐ and RF‐spoiled gradient echo sequences

AbstractPurposeThe complex signal decay during the transient FLASH MRI readout can lead to artifacts in magnitude and phase images. We show that target‐driven optimization of individual RF flip angles and phases can realize near‐ideal signal behavior and mitigate artifacts.MethodsThe differentiable end‐to‐end optimization framework MR‐zero is used to optimize RF trains of the FLASH sequence. We focus herein on minimizing deviations from the ideally spoiled signal by using a mono‐exponential Look–Locker target. We first obtain the transient FLASH signal decay substructure, and then minimize the deviation to the Look–Locker decay by optimizing the individual (i) flip angles, (ii) RF phases, and (iii) flip angles and RF phases. Comparison between measurement and simulation is performed using Pulseq in 1D and 2D.ResultsWe were able to reproduce the complex substructure of the transient FLASH signal decay. All three optimization objectives can bring the real FLASH signal closer to the ideal case, with best results when both flip angles and RF phases are adjusted jointly. This solution outperformed all tested conventional quadratic RF cyclings in terms of (i) matching the Look–Locker target signal, (ii) phase stability, (iii) point spread functions ideality, (iv) robustness against parameter changes, and (v) magnitude and phase image quality. Other target functions for the signal could as well be realized, yet their response is not as general as for the Look–Locker target and needs to be optimized for a specific context.ConclusionIndividual flip angle and RF phase optimization improves the transient signal decay of FLASH MRI sequences.

MRI denoising with a non-blind deep complex-valued convolutional neural network.

MR images with high signal-to-noise ratio (SNR) provide more diagnostic information. Various methods for MRI denoising have been developed, but the majority of them operate on the magnitude image and neglect the phase information. Therefore, the goal of this work is to design and implement a complex-valued convolutional neural network (CNN) for MRI denoising. A complex-valued CNN incorporating the noise level map (non-blind DnCNN) was trained with ground truth and simulated noise-corrupted image pairs. The proposed method was validated using both simulated and in vivo data collected from low-field scanners. Its denoising performance was quantitively and qualitatively evaluated, and it was compared with the real-valued CNN and several other algorithms. For the simulated noise-corrupted testing dataset, the complex-valued models had superior normalized root-mean-square error, peak SNR, structural similarity index, and phase ABSD. By incorporating the noise level map, the non-blind DnCNN showed better performance in dealing with spatially varying parallel imaging noise. For in vivo low-field data, the non-blind DnCNN significantly improved the SNR and visual quality of the image. The proposed non-blind DnCNN provides an efficient and effective approach for MRI denoising. This is the first application of non-blind DnCNN to medical imaging. The method holds the potential to enable improved low-field MRI, facilitating enhanced diagnostic imaging in under-resourced areas.

Improving -inhomogeneity tolerance by resolving non-bijection in MP2RAGE R1 mapping: A 2D look-up table approach demonstrated at 3 T.

Regarding in vivo, robust longitudinal relaxation rate (R1) mapping, the goal of the present paper is two-fold. First, to verify that non-bijective mapping in magnetization-prepared 2 rapid gradient echo (MP2RAGE) imaging can be resolved through a two-dimensional look-up table approach. Second, that the expanded parameter space from this can be used to improve -inhomogeneity tolerance without other prerequisites. By deriving a second contrast from the magnitude images of the MP2RAGE acquisition, ambiguities in the original MP2RAGE image resulting from non-bijective transfer curves can be resolved. Such ambiguities may occur when protocols are optimized, such as for higher -inhomogeneity tolerance. A 2D look-up table approach combines the available information to resolve these ambiguities during mapping. At 3 T, we acquired MP2RAGE images with standard acquisition parameters and (non-bijective) parameters optimized for -inhomogeneity tolerance. From 3 subjects across multiple sessions, we assessed the -inhomogeneity tolerance through excitation-pulse amplitude scalings. The R1 maps resulting from the -optimized protocols showed greatly reduced effects across images but without additional scanner time. Meanwhile, these maps could only successfully be derived by a 2D look-up table approach. We show that it is possible to optimize for -inhomogeneity tolerance in MP2RAGE through sequence-parameter settings, while still successfully estimating the R1 map with a two-dimensional look-up table approach. This without the need for an additional map. The increased parameter space enabled by the two-dimensional look-up table approach may further be used to adjust MP2RAGE acquisitions for improved scan times, signal-to-noise ratio, and/or contrast-to-noise ratio.

Automatic segmentation and diameter measurement of deep medullary veins.

As one of the pathogenic factors of cerebral small vessel disease, venous collagenosis may result in the occlusion or stenosis of deep medullary veins (DMVs). Although numerous DMVs can be observed in susceptibility-weighted MRI images, their diameters are usually smaller than the MRI resolution, making it difficult to segment them and quantify their sizes. We aim to automatically segment DMVs and measure their diameters from gradient-echo images. A neural network model was trained for DMV segmentation based on the gradient-echo magnitude and phase images of 20 subjects at 7 T. The diameters of DMVs were obtained by fitting measured complex images with model images that accounted for the DMV-induced magnetic field and point spread function. A phantom study with graphite rods of different diameters was conducted to validate the proposed method. Simulation was carried out to evaluate the voxel-size dependence of measurement accuracy for a typical DMV size. The automatically segmented DMV masks had Dice similarity coefficients of 0.68 ± 0.03 (voxel level) and 0.83 ± 0.04 (cluster level). The fitted graphite-rod diameters closely matched their true values. In simulation, the fitted diameters closely matched the true value when voxel size was ≤ 0.45 mm, and 92.2% of DMVs had diameters between 90 μm and 200 μm with a peak at about 120 μm, which agreed well with an earlier ex vivo report. The proposed methods enabled efficient and quantitative study of DMVs, which may help illuminate the role of DMVs in the etiopathogenesis of cerebral small vessel disease.

Complex-valued image reconstruction for compressed sensing MRI using hierarchical constraint

In Magnetic Resonance Imaging (MRI), the sequential acquisition of raw complex-valued image data in Fourier space, also known as k-space, results in extended examination times. To speed up the MRI scans, k-space data are usually undersampled and processed using numerical techniques such as compressed sensing (CS). While the majority of CS-MRI algorithms primarily focus on magnitude images due to their significant diagnostic value, the phase components of complex-valued MRI images also hold substantial importance for clinical diagnosis, including neurodegenerative diseases. In this work, complex-valued MRI reconstruction is studied with a focus on the simultaneous reconstruction of both magnitude and phase images. The proposed algorithm is based on the nonsubsampled contourlet transform (NSCT) technique, which offers shift invariance in images. Instead of directly transforming the complex-valued image into the NSCT domain, we introduce a wavelet transform within the NSCT domain, reducing the size of the sparsity of coefficients. This two-level hierarchical constraint (HC) enforces sparse representation of complex-valued images for CS-MRI implementation. The proposed HC is seamlessly integrated into a proximal algorithm simultaneously. Additionally, to effectively minimize the artifacts caused by sub-sampling, thresholds related to different sub-bands in the HC are applied through an alternating optimization process. Experimental results show that the novel method outperforms existing CS-MRI techniques in phase-regularized complex-valued image reconstructions.

Deep learning-based multi-frequency denoising for myocardial perfusion SPECT

BackgroundDeep learning (DL)-based denoising has been proven to improve image quality and quantitation accuracy of low dose (LD) SPECT. However, conventional DL-based methods used SPECT images with mixed frequency components. This work aims to develop an integrated multi-frequency denoising network to further enhance LD myocardial perfusion (MP) SPECT denoising.MethodsFifty anonymized patients who underwent routine 99mTc-sestamibi stress SPECT/CT scans were retrospectively recruited. Three LD datasets were obtained by reducing the 10 s acquisition time of full dose (FD) SPECT to be 5, 2 and 1 s per projection based on the list mode data for a total of 60 projections. FD and LD projections were Fourier transformed to magnitude and phase images, which were then separated into two or three frequency bands. Each frequency band was then inversed Fourier transformed back to the image domain. We proposed a 3D integrated attention-guided multi-frequency conditional generative adversarial network (AttMFGAN) and compared with AttGAN, and separate AttGAN for multi-frequency bands denoising (AttGAN-MF).The multi-frequency FD and LD projections of 35, 5 and 10 patients were paired for training, validation and testing. The LD projections to be tested were separated to multi-frequency components and input to corresponding networks to get the denoised components, which were summed to get the final denoised projections. Voxel-based error indices were measured on the cardiac region on the reconstructed images. The perfusion defect size (PDS) was also analyzed.ResultsAttGAN-MF and AttMFGAN have superior performance on all physical and clinical indices as compared to conventional AttGAN. The integrated AttMFGAN is better than AttGAN-MF. Multi-frequency denoising with two frequency bands have generally better results than corresponding three-frequency bands methods.ConclusionsAttGAN-MF and AttMFGAN are promising to further improve LD MP SPECT denoising.

Susceptibility-Weighted Imaging for Differentiating Intracranial Calcification and Hemosiderin: A Comparative Study with CT

Background: Magnetic resonance imaging (MRI) is an essential diagnostic modality for evaluating various tissue types, but distinguishing between substances such as calcification and hemosiderin remains challenging due to their similar appearances on conventional MRI sequences. Susceptibility-weighted imaging (SWI) improves tissue contrast by incorporating phase information, potentially enhancing the accuracy of differentiating these substances. This study aims to assess the effectiveness of SWI, specifically the filtered phase image, in comparison with computed tomography (CT), which is considered the gold standard for calcification detection. Methods: A retrospective analysis was conducted on 35 patients, aged 50-80 years, who presented with confirmed cases of intracranial calcification or hemosiderin deposits. MRI was performed on a 1.5-tesla Siemens Essenza Magnetom using SWI sequences, including both phase and magnitude images. CT scans were obtained using a GE Optima 660 scanner. The study focused on 17 cases of calcification identified by both SWI and CT, and 18 cases of hemosiderin deposits detected by SWI. Statistical analysis was performed using Pearson correlation coefficients to evaluate the relationship between MRI and CT measurements. Results: The study revealed a strong correlation between CT and SWI in detecting calcifications, with a significant agreement in the measurement of calcification size. However, while phase and magnitude images demonstrated a strong correlation with one another, the filtered phase image did not show a significant correlation with hemosiderin size, suggesting limitations in its application for detecting hemosiderin deposits. CT scans confirmed calcification with Hounsfield units exceeding 100, while hemosiderin deposits were more effectively identified through SWI, which provided enhanced contrast without the use of ionizing radiation. Conclusion: SWI, particularly the filtered phase image, is a reliable and sensitive tool for detecting intracranial calcification, offering high accuracy compared to CT. While SWI proves effective in distinguishing calcification, further research is needed to optimize its use for detecting hemosiderin deposits. This study underscores the non-ionizing advantages of SWI over CT, advocating for its potential as a safer and equally effective alternative in clinical imaging.

Enhancing organ and vascular contrast in preclinical ultra-low field MRI using superparamagnetic iron oxide nanoparticles

Superparamagnetic iron oxide nanoparticles (SPIONs) are characterized by their exceptional susceptibility and relaxivity at ultra-low field (ULF) regimes, make them a promising contrast agent (CA) for ULF MRI. Despite their distinct advantages, the translation of these properties into clinically valuable image contrast in ULF MRI remains underexplored. In this study, we investigate the use of SPIONs to generate in vivo MRI contrast at 6.5 mT within the organs and vascular system of rodents. This investigation includes comprehensive SPION characterization and phantom imaging experiments to validate the utility of SPIONs to produce positive image contrast and to facilitate phase-sensitive imaging at ULF. Optimized balanced steady-state free precession (bSSFP) and spoiled gradient echo (SPGR) MRI sequences are used to generate in vivo contrast by leveraging the distinctive properties of SPIONs at ULF. Imaging studies in rodents reveal positive organ contrast attainable in magnitude images, and MRI phase maps can be used to visualize the vascular system. This work demonstrates the effectiveness of SPIONs in enhancing preclinical organ and vascular imaging at ULF; it bridges the gap between the study of the distinctive physical properties of SPIONs and the demonstration of in vivo image contrast.

Deep-learning-based image reconstruction with limited data: generating synthetic raw data using deep learning

ObjectDeep learning has shown great promise for fast reconstruction of accelerated MRI acquisitions by learning from large amounts of raw data. However, raw data is not always available in sufficient quantities. This study investigates synthetic data generation to complement small datasets and improve reconstruction quality.Materials and methodsAn adversarial auto-encoder was trained to generate phase and coil sensitivity maps from magnitude images, which were combined into synthetic raw data.On a fourfold accelerated MR reconstruction task, deep-learning-based reconstruction networks were trained with varying amounts of training data (20 to 160 scans). Test set performance was compared between baseline experiments and experiments that incorporated synthetic training data.ResultsTraining with synthetic raw data showed decreasing reconstruction errors with increasing amounts of training data, but importantly this was magnitude-only data, rather than real raw data. For small training sets, training with synthetic data decreased the mean absolute error (MAE) by up to 7.5%, whereas for larger training sets the MAE increased by up to 2.6%.DiscussionSynthetic raw data generation improved reconstruction quality in scenarios with limited training data. A major advantage of synthetic data generation is that it allows for the reuse of magnitude-only datasets, which are more readily available than raw datasets.

Ultrashort echo time magnetic resonance elastography for quantification of the mechanical properties of short T2 tissues via optimal control-based radiofrequency pulses.

The aim of the current study is to demonstrate the feasibility of radiofrequency (RF) pulses generated via an optimal control (OC) algorithm to perform magnetic resonance elastography (MRE) and quantify the mechanical properties of materials with very short transverse relaxation times (T2 < 5 ms) for the first time. OC theory applied to MRE provides RF pulses that bring isochromats from the equilibrium state to a fixed target state, which corresponds to the phase pattern of a conventional MRE acquisition. Such RF pulses applied with a constant gradient allow to simultaneously perform slice selection and motion encoding in the slice direction. Unlike conventional MRE, no additional motion-encoding gradients (MEGs) are needed, enabling shorter echo times. OC pulses were implemented both in turbo spin echo (OC rapid acquisition with refocused echoes [RARE]) and ultrashort echo time (OC UTE) sequences to compare their motion-encoding efficiency with the conventional MEG encoding (classical MEG MRE). MRE experiments were carried out on agar phantoms with very short T2 values and on an ex vivo bovine tendon. Magnitude images, wave field images, phase-to-noise ratio (PNR), and shear storage modulus maps were compared between OC RARE, OC UTE, and classical MEG MRE in samples with different T2 values. Shear storage modulus values of the agar phantoms were in agreement with values found in the literature, and that of the bovine tendon was corroborated with rheometry measurements. Only the OC sequences could encode motion in very short T2 samples, and only OC UTE sequences yielded magnitude images enabling proper visualization of short T2 samples and tissues. The OC UTE sequence produced the best PNRs, demonstrating its ability to perform anatomical and mechanical characterization. Its success warrants in vivo confirmation in further studies.

Quantitative susceptibility mapping based basal ganglia segmentation via AGSeg: leveraging active gradient guiding mechanism in deep learning.

With better visual contrast and the ability for magnetic susceptibility quantification analysis, quantitative susceptibility mapping (QSM) has emerged as an important magnetic resonance imaging (MRI) method for basal ganglia studies. Precise segmentation of basal ganglia is a prerequisite for quantification analysis of tissue magnetic susceptibility, which is crucial for subsequent disease diagnosis and surgical planning. The conventional method of localizing and segmenting basal ganglia heavily relies on layer-by-layer manual annotation by experts, resulting in a tedious amount of workload. Although several morphology registration and deep learning based methods have been developed to automate segmentation, the voxels around the nuclei boundary remain a challenge to distinguish due to insufficient tissue contrast. This paper proposes AGSeg, an active gradient guidance-based susceptibility and magnitude information complete (MIC) network for real-time and accurate basal ganglia segmentation. Various datasets, including clinical scans and data from healthy volunteers, were collected across multiple centers with different magnetic field strengths (3T/5T/7T), with a total of 210 three-dimensional (3D) susceptibility measurements. Manual segmentations following fixed rules for anatomical borders annotated by experts were used as ground truth labels. The proposed network took QSM maps and Magnitude images as two individual inputs, of which the features are selectively enhanced in the proposed magnitude information complete (MIC) module. AGSeg utilized a dual-branch architecture, with Seg-branch aiming to generate a proper segmentation map and Grad-branch to reconstruct the gradient map of regions of interest (ROIs). With the support of the newly designed active gradient module (AGM) and gradient guiding module (GGM), the Grad-branch provided attention guidance for the Seg-branch, facilitating it to focus on the boundary of target nuclei. Ablation studies were conducted to assess the functionality of the proposed modules. Significant performance decrement was observed after ablating relative modules. AGSeg was evaluated against several existing methods on both healthy and clinical data, achieving an average Dice similarity coefficient (DSC) =0.874 and average 95% Hausdorff distance (HD95) =2.009. Comparison experiments indicated that our model had superior performance on basal ganglia segmentation and better generalization ability over existing methods. The AGSeg outperformed all implemented comparison deep learning algorithms with average DSC enhancement ranging from 0.036 to 0.074. The current work integrates a deep learning-based method into automated basal ganglia segmentation. The high processing speed and segmentation robustness of AGSeg contribute to the feasibility of future surgery planning and intraoperative navigation. Experiments show that leveraging active gradient guidance mechanisms and magnitude information completion can facilitate the segmentation process. Moreover, this approach also offers a portable solution for other multi-modality medical image segmentation tasks.

Unveiling metabolic patterns in dementia: Insights from high-resolution quantitative blood-oxygenation-level-dependent MRI.

Oxygen extraction fraction (OEF) and deoxyhemoglobin (DoHb) levels reflect variations in cerebral oxygen metabolism in demented patients. Delineating the metabolic profiles evident throughout different phases of dementia necessitates an integrated analysis of OEF and DoHb levels. This is enabled by leveraging high-resolution quantitative blood oxygenation level dependent (qBOLD) analysis of magnitude images obtained from a multi-echo gradient-echo MRI (mGRE) scan performed on a 3.0 Tesla scanner. Achieving superior spatial resolution in qBOLD necessitates the utilization of an mGRE scan with only four echoes, which in turn limits the number of measurements compared to the parameters within the qBOLD model. Consequently, it becomes imperative to discard non-essential parameters to facilitate further analysis. This process entails transforming the qBOLD model into a format suitable for fitting the log-magnitude difference (L-MDif) profiles of the four echo magnitudes present in each brain voxel. In order to bolster spatial specificity, the log-difference qBOLD model undergoes refinement into a representative form, termed as r-qBOLD, particularly when applied to class-averaged L-MDif signals derived through k-means clustering of L-MDif signals from all brain voxels into a predetermined number of clusters. The agreement between parameters estimated using r-qBOLD for different cluster sizes is validated using Bland-Altman analysis, and the model's goodness-of-fit is evaluated using a -test. Retrospective MRI data of Alzheimer's disease (AD), mild cognitive impairment (MCI), and non-demented patients without neuropathological disorders, pacemakers, other implants, or psychiatric disorders, who completed a minimum of three visits prior to MRI enrolment, are utilized for the study. Utilizing a cohort comprising 30 demented patients aged 65-83 years in stages 4-6 representing mild, moderate, and severe stages according to the clinical dementia rating (CDR), matched with an age-matched non-demented control group of 18 individuals, we conducted joint observations of OEF and DoHb levels estimated using r-qBOLD. The observations elucidate metabolic signatures in dementia based on OEF and DoHb levels in each voxel. Our principal findings highlight the significance of spatial patterns of metabolic profiles (metabolic patterns) within two distinct regimes: OEF levels exceeding the normal range (S1-regime), and OEF levels below the normal range (S2-regime). The S1-regime, accompanied by low DoHb levels, predominantly manifests in fronto-parietal and perivascular regions with increase in dementia severity. Conversely, the S2-regime, accompanied by low DoHb levels, is observed in medial temporal (MTL) regions. Other regions with abnormal metabolic patterns included the orbitofrontal cortex (OFC), medial-orbital prefrontal cortex (MOPFC), hypothalamus, ventro-medial prefrontal cortex (VMPFC), and retrosplenial cortex (RSP). Dysfunction in the OFC and MOPFC indicated cognitive and emotional impairment, while hypothalamic involvement potentially indicated preclinical dementia. Reduced metabolic activity in the RSP suggested early-stage AD related functional abnormalities. Integrated analysis of OEF and DoHb levels using r-qBOLD reveals distinct metabolic signatures across dementia phases, highlighting regions susceptible to neuronal loss, vascular involvement, and preclinical indicators.

Radial Undersampled MRI Reconstruction Using Deep Learning With Mutual Constraints Between Real and Imaginary Components of K-Space.

The deep learning method is an efficient solution for improving the quality of undersampled magnetic resonance (MR) image reconstruction while reducing lengthy data acquisition. Most deep learning methods neglect the mutual constraints between the real and imaginary components of complex-valued k-space data. In this paper, a new complex-valued convolutional neural network, namely, Dense-U-Dense Net (DUD-Net), is proposed to interpolate the undersampled k-space data and reconstruct MR images. The proposed network comprises dense layers, U-Net, and other dense layers in sequence. The dense layers are used to simulate the mutual constraints between real and imaginary components, and U-Net performs feature sparsity and interpolation estimation for k-space data. Two MRI datasets were used to evaluate the proposed method: brain magnitude-only MR images and knee complex-valued k-space data. Several operations were conducted for data preprocessing. First, the complex-valued MR images were synthesized by phase modulation on magnitude-only images. Second, a radial trajectory based on the golden angle was used for k-space undersampling, whereby a reversible normalization method was proposed to balance the distribution of positive and negative values in k-space data. The optimal performance of DUD-Net was demonstrated based on a quantitative evaluation of inter-method and intra-method comparisons. When compared with other methods, significant improvements were achieved, PSNRs were increased by 10.78 and 5.74dB, whereas RMSEs were decreased by 71.53% and 30.31% for magnitude and phase image, respectively. It is concluded that DUD-Net significantly improves the performance of MR image reconstruction.

Tumor Stiffness Measurement at Multifrequency MR Elastography to Predict Lymphovascular Space Invasion in Endometrial Cancer.

Background Pathologic lymphovascular space invasion (LVSI) is associated with poor outcome in endometrial cancer. Its relationship with tumor stiffness, which can be measured with use of MR elastography, has not been extensively explored. Purpose To assess whether MR elastography-based mechanical characteristics can aid in the noninvasive prediction of LVSI in patients with endometrial cancer. Materials and Methods This prospective study included consecutive adult patients with a suspected uterine tumor who underwent MRI and MR elastography between October 2022 and July 2023. A region of interest delineated on T2-weighted magnitude images was duplicated on MR elastography images and used to calculate c (stiffness in meters per second) and φ (viscosity in radians) values. Pathologic assessment of hysterectomy specimens for LVSI served as the reference standard. Data were compared between LVSI-positive and -negative groups with use of the Mann-Whitney U test. Multivariable logistic regression was used to determine variables associated with LVSI positivity and develop diagnostic models for predicting LVSI. Model performance was assessed with use of area under the receiver operating characteristic curve (AUC) and compared using the DeLong test. Results A total of 101 participants were included, 72 who were LVSI-negative (median age, 53 years [IQR, 48-62 years]) and 29 who were LVSI-positive (median age, 54 years [IQR, 49-60 years]). The tumor stiffness in the LVSI-positive group was higher than in the LVSI-negative group (median, 4.1 m/sec [IQR, 3.2-4.6 m/sec] vs 2.2 m/sec [IQR, 2.0-2.8 m/sec]; P < .001). Tumor volume, cancer antigen 125 level, and tumor stiffness were associated with LVSI positivity (adjusted odds ratio range, 1.01-9.06; P range, <.001-.04). The combined model (AUC, 0.93) showed better performance for predicting LVSI compared with clinical-radiologic model (AUC, 0.77; P = .003) and similar performance to the MR elastography-based model (AUC, 0.89; P = .06). Conclusion The addition of tumor stiffness as measured at MR elastography into a clinical-radiologic model improved prediction of LVSI in patients with endometrial cancer. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Ehman in this issue.

Impact of phase-correction of late-gadolinium enhancement images on quantification of scar metrics and in-silico VT-modelling in patients with an ischemic cardiomyopathy

Abstract Background Late gadolinium enhancement (LGE) is used to perform tissue characterization and has become the cornerstone to guide treatment of ventricular tachyarrhythmia (VT). Traditionally, LGE acquisition is reconstructed in two distinct forms: non-phase corrected magnitude (MIR) and phase-corrected inversion recovery (PSIR) images. However, the influence of MIR vs. PSIR on the quantification of scar metrics and the repercussion on estimating myocardial arrhythmogenicity have not been comprehensively explored. Aim This study investigated the influence of PSIR and MIR LGE on the quantification of scar metrics in a randomly selected cohort of patients with ischemic cardiomyopathy who had undergone imaging prior to ICD implantation and had received device therapy during follow-up. In addition, this study assessed how these image variations affect functional estimation of arrhythmogenicity using computational modelling. Methods ADAS 3D LV was used to generate scar maps from 2D PSIR and MIR LGE images in 10 patients. The extent of variation in scar-characteristics was evaluated by quantifying the borderzone (BZ), scar core, and conduction corridors. Electrical simulations were performed on the patient-specific anatomical models using the Virtual Induction and Treatment of Arrhythmias (VITA) framework. VITA was used to assess the electrical vulnerability of potential reentrant channels and provided the number of inducible VTs and their corresponding round-trip-time (RTT), an equivalent of VT cycle-length. Results Scar quantification using PSIR LGE resulted in a significantly larger BZ (22.1±23.7g vs. 6.2±2.7g, p &amp;lt;.001) and scar core (43.5±21.5g vs. 10.5±4g, p =.047). Furthermore, both the absolute number (6.8±4 vs. 3.4±1.5, p = .031) and weight (13±13.1g vs. 1.4±0.8g, p =.018) of conduction corridors were also higher using PSIR. VTs were inducible in 8 PSIR-models and 7 MIR-models. Despite large differences in scar characteristics, VITA metrics were not significantly different between PSIR and MIR derived models. Although, the absolute number of inducible VTs were larger for PSIR (6.2±5.4 vs. 3.4±3.1), non-parametric testing using Wilcoxon signed-rank test did not demonstrate a significant difference (Z = -1.334, p =.139). A comparable, non-significant difference was observed for mean- (117±78 vs. 79±60, p =.257) and max-RTT (181±141 vs. 120±103, p =.294) between PSIR and MIR derived models. Conclusion PSIR resulted in significantly larger BZ, scar core area, and conduction corridors. While the absolute, VITA-derived estimation of myocardial arrhythmogenicity, including inducible VTs and RTT, were higher in the PSIR-based quantification, these differences did not reach statistical significance. These results highlight important differences between two clinically applied image reconstruction techniques for the assessment of arrhythmogenic substrate, emphasizing the need for additional investigation and refinement of scar quantification techniques.

Cross‐attention neural network for land cover change detection with remote sensing images

AbstractLand cover change detection (LCCD) with remote sensing images (RSIs) is important for observing the land cover change of the Earth's surface. Considering the insufficient performance of the traditional self‐attention mechanism used in a neural network to smoothen the noise of LCCD with RSIs, in this study a novel cross‐attention neural network (CANN) was proposed for improving the performance of LCCD with RSIs. In the proposed CANN, a cross‐attention mechanism was achieved by employing another temporal image to enhance attention performance and improve detection accuracies. First, a feature difference module was embedded in the backbone of the proposed CANN to generate a change magnitude image and guide the learning progress. A self‐attention module based on the cross‐attention mechanism was then proposed and embedded in the encoder of the proposed network to make the network pay attention to the changed area. Finally, the encoded features were decoded to obtain binary change detection with the ArgMax function. Compared with five methods, the experimental results based on six pairs of real RSIs well demonstrated the feasibility and superiority of the proposed network for achieving LCCD with RSIs. For example, the improvement for overall accuracy for the six pairs of real RSIs improved by our proposed approach is about 0.72–2.56%.

Robust multishot diffusion-weighted imaging of the abdomen with region-based shot rejection.

Diffusion-weighted (DW) imaging provides a useful clinical contrast, but is susceptible to motion-induced dephasing caused by the application of strong diffusion gradients. Phase navigators are commonly used to resolve shot-to-shot motion-induced phase in multishot reconstructions, but poor phase estimates result in signal dropout and Apparent Diffusion Coefficient (ADC) overestimation. These artifacts are prominent in the abdomen, a region prone to involuntary cardiac and respiratory motion. To improve the robustness of DW imaging in the abdomen, region-based shot rejection schemes that selectively weight regions where the shot-to-shot phase is poorly estimated were evaluated. Spatially varying weights for each shot, reflecting both the accuracy of the estimated phase and the degree of subvoxel dephasing, were estimated from the phase navigator magnitude images. The weighting was integrated into a multishot reconstruction using different formulations and phase navigator resolutions and tested with different phase navigator resolutions in multishot DW-echo Planar Imaging acquisitions of the liver and pancreas, using conventional monopolar and velocity-compensated diffusion encoding. Reconstructed images and ADC estimates were compared qualitatively. The proposed region-based shot rejection reduces banding and signal dropout artifacts caused by physiological motion in the liver and pancreas. Shot rejection allows conventional monopolar diffusion encoding to achieve median ADCs in the pancreas comparable to motion-compensated diffusion encoding, albeit with a greater spread of ADCs. Region-based shot rejection is a linear reconstruction that improves the motion robustness of multi-shot DWI and requires no sequence modifications.

Interscanner reproducibility of volumetric quantitative susceptibility mapping about cerebral subcortical gray nuclei at different MR vendors with the same magnetic strength.

Quantitative susceptibility mapping (QSM) technique was a new quantitative magnetic resonance imaging technique to evaluate the cerebral iron deposition in clinical practice. The current study was aimed to investigate the reproducibility of the volumetric susceptibility value of the subcortical gray nuclei at two different MR vendor with the same magnetic strength. Cerebral magnitude and phase images of 21 normal subjects were acquired from a 3D multiecho enhanced gradient recalled echo sequence at two different 3.0T MR scanner, and then the magnetic susceptibility images were generated by STI software. The brain structural images were coregistered with magnitude images and generated the normalized parameters, and then generated the normalized susceptibility images. The subcortical gray nuclei template was applied to extract the volumetric susceptibility value of the target nuclei. ICC value (95% CI) of the caudate, putamen and GP were 0.847 (0.660-0.935), 0.848 (0.663-0.935) and 0.838 (0.643-0.931), respectively. The ICC value of the thalamus was 0.474 (0.064-0.747). Ninety-five point two percent (20/21) of the difference points of the susceptibility located between the 95% LA for the caudate at the two different 3.0T MR scanner, while the less than 95% of the difference points of the susceptibility value located between the 95% LA for the putamen, globus pallidus and thalamus. The current study identified that the caudate had the stable reproducibility of the magnetic susceptibility value, and the other basal ganglion nuclei should be cautious for the quantitative evaluation of the magnetic susceptibility value at different 3.0T MR scanner.

Quantifying cerebral microbleeds using quantitative susceptibility mapping from magnetization-prepared rapid gradient-echo.

T1-weighted magnetization-prepared rapid gradient-echo (MPRAGE) is commonly included in brain studies for structural imaging using magnitude images; however, its phase images can provide an opportunity to assess microbleed burden using quantitative susceptibility mapping (QSM). This potential application for MPRAGE-based QSM was evaluated using in vivo and simulated measurements. Possible factors affecting image quality were also explored. Detection sensitivity was evaluated against standard multiecho gradient echo (MEGE) QSM using 3-T in vivo data of 15 subjects with a combined total of 108 confirmed microbleeds. The two methods were compared based on the microbleed size and susceptibility measurements. In addition, simulations explored the detection sensitivity of MPRAGE-QSM at different representative magnetic field strengths and echo times using microbleeds of different size, susceptibility, and location. Results showed that in vivo microbleeds appeared to be smaller (× 0.54) and of higher mean susceptibility (× 1.9) on MPRAGE-QSM than on MEGE-QSM, but total susceptibility estimates were in closer agreement (slope: 0.97, r2: 0.94), and detection sensitivity was comparable. In simulations, QSM at 1.5 T had a low contrast-to-noise ratio that obscured the detection of many microbleeds. Signal-to-noise ratio (SNR) levels at 3 T and above resulted in better contrast and increased detection. The detection rates for microbleeds of minimum one-voxel diameter and 0.4-ppm susceptibility were 0.55, 0.80, and 0.88 at SNR levels of 1.5, 3, and 7 T, respectively. Size and total susceptibility estimates were more consistent than mean susceptibility estimates, which showed size-dependent underestimation. MPRAGE-QSM provides an opportunity to detect and quantify the size and susceptibility of microbleeds of at least one-voxel diameter at B0of 3 T or higher with no additional time cost, when standard T2*-weighted images are not available or have inadequate spatial resolution. The total susceptibility measure is more robust against sequence variations and might allow combining data from different protocols.