Plug-and-Play latent feature editing for orientation-adaptive quantitative susceptibility mapping neural networks.

Quantitative susceptibility mapping (QSM) is a post-processing technique for deriving tissue magnetic susceptibility distribution from MRI phase measurements. Deep learning (DL) algorithms hold great potential for solving the ill-posed QSM reconstruction problem. However, a significant challenge facing current DL-QSM approaches is their limited adaptability to magnetic dipole field orientation variations during training and testing. In this work, we propose a novel Orientation-Adaptive Latent Feature Editing (OA-LFE) module to learn the encoding of acquisition orientation vectors and seamlessly integrate them into the latent features of deep networks. Importantly, it can be directly Plug-and-Play (PnP) into various existing DL-QSM architectures, enabling reconstructions of QSM from arbitrary magnetic dipole orientations. Its effectiveness is demonstrated by combining the OA-LFE module into our previously proposed phase-to-susceptibility single-step instant QSM (iQSM) network, which was initially tailored for pure-axial acquisitions. The proposed OA-LFE-empowered iQSM, which we refer to as iQSM+, is trained in a simulated-supervised manner on a specially-designed simulation brain dataset. Comprehensive experiments are conducted on simulated and in vivo human brain datasets, encompassing subjects ranging from healthy individuals to those with pathological conditions. These experiments involve various MRI platforms (3T and 7T) and aim to compare our proposed iQSM+ against several established QSM reconstruction frameworks, including the original iQSM. The iQSM+ yields QSM images with significantly improved accuracies and mitigates artifacts, surpassing other state-of-the-art DL-QSM algorithms. The PnP OA-LFE module's versatility was further demonstrated by its successful application to xQSM, a distinct DL-QSM network for dipole inversion. In conclusion, this work introduces a new DL paradigm, allowing researchers to develop innovative QSM methods without requiring a complete overhaul of their existing architectures.

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 B0  of 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.

MR Susceptibility Separation for Quantifying Lesion Paramagnetic and Diamagnetic Evolution in Relapsing-Remitting Multiple Sclerosis.

BACKGROUND: Multiple sclerosis (MS) lesion evolution may involve changes in diamagnetic myelin and paramagnetic iron. Conventional quantitative susceptibility mapping (QSM) can provide net susceptibility distribution, but not the discrete paramagnetic and diamagnetic components. PURPOSE: To apply susceptibility separation (χ separation) to follow lesion evolution in MS with comparison to R2 */R2 ' /QSM. STUDY TYPE: Longitudinal, prospective. SUBJECTS: Twenty relapsing-remitting MS subjects (mean age: 42.5 ± 9.4 years, 13 females; mean years of symptoms: 4.3 ± 1.4 years). FIELD STRENGTH/SEQUENCE: Three-dimensional multiple echo gradient echo (QSM and R2 * mapping), two-dimensional dual echo fast spin echo (R2 mapping), T2 -weighted fluid attenuated inversion recovery, and T1-weighted magnetization prepared gradient echo sequences at 3 T. ASSESSMENT: Data were analyzed from two scans separated by a mean interval of 14.4 ± 2.0 months. White matter lesions on fluid-attenuated inversion recovery were defined by an automatic pipeline, then manually refined (by ZZ/AHW, 3/25 years' experience in MRI), and verified by a radiologist (MN, 25 years' experience in MS). Susceptibility separation yielded the paramagnetic and diamagnetic susceptibility content of each voxel. Lesions were classified into four groups based on the variation of QSM/R2 * or separated into positive/negative components from χ separation. STATISTICAL TESTS: Two-sample paired t tests for assessment of longitudinal differences. Spearman correlation coefficients to assess associations between χ separation and R2 */R2 ' /QSM. Significant level: P < 0.005. RESULTS: A total of 183 lesions were quantified. Categorizing lesions into groups based on χ separation demonstrated significant annual changes in QSM//R2 */R2 ' . When lesions were grouped based on changes in QSM and R2 *, both changing in unison yielded a significant dominant paramagnetic variation and both opposing yielded a dominant diamagnetic variation. Significant Spearman correlation coefficients were found between susceptibility-sensitive MRI indices and χ separation. DATA CONCLUSION: Susceptibility separation changes in MS lesions may distinguish and quantify paramagnetic and diamagnetic evolution, potentially providing additional insight compared to R2 * and QSM alone. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY: Stage 2.

Simultaneous time-of-flight MR angiography and quantitative susceptibility mapping with key time-of-flight features.

A technique for combined time-of-flight (TOF) MR angiography (MRA) and quantitative susceptibility mapping (QSM) was developed with key features of standard three-dimensional (3D) TOF acquisitions, including multiple overlapping thin slab acquisition (MOTSA), ramped RF excitation, and venous saturation. The developed triple-echo 3D TOF-QSM sequence enabled TOF-MRA, susceptibility-weighted imaging (SWI), QSM, and R2* mapping. The effects of ramped RF, resolution, flip angle, venous saturation, and MOTSA were studied on QSM. Six volunteers were scanned at 3 T with the developed sequence, conventional TOF-MRA, and conventional SWI. Quantitative comparison of susceptibility values on QSM and normalized arterial and venous vessel-to-background contrasts on TOF and SWI were performed. The ramped RF excitation created an inherent phase variation in the raw phase. A generic correction factor was computed to remove the phase variation to obtain QSM without artifacts from the TOF-QSM sequence. No statistically significant difference was observed between the developed and standard QSM sequence for susceptibility values. However, maintaining standard TOF features led to compromises in signal-to-noise ratio for QSM and SWI, arising from the use of MOTSA rather than one large 3D slab, higher TOF spatial resolution, increased TOF background suppression due to larger flip angles, and reduced venous signal from venous saturation. In terms of vessel contrast, veins showed higher normalized contrast on SWI derived from TOF-QSM than the standard SWI sequence. While fast flowing arteries had reduced contrast compared with standard TOF-MRA, no statistical difference was observed for slow flowing arteries. Arterial contrast differences largely arise from the longer TR used in TOF-QSM over standard TOF-MRA to accommodate additional later echoes for SWI. In conclusion, although the sequence has a longer TR and slightly lower arterial contrast, provided an adequate correction is made for ramped RF excitation effects on phase, QSM may be performed from a multiecho sequence that includes all key TOF features, thus enabling simultaneous TOF-MRA, SWI, QSM, and R2* map computation.

Repurposing of the PET Probe Prototype PiB for Label and Radiation-Free CEST MRI Molecular Imaging of Amyloid.

Positron emission tomography (PET) probes are specific and sensitive while suffering from radiation risk. It is worthwhile to explore the chemical emission saturation transfer (CEST) effects of the probe prototypes and repurpose them for CEST imaging to avoid radiation. In this study, we used 11C-PiB as an example of a PET probe for detecting amyloid and tested the feasibility of repurposing this PET probe prototype, PiB, for CEST imaging. After optimizing the parameters through preliminary phantom experiments, we used APP/PS1 transgenic mice and age-matched C57 mice for in vivo CEST magnetic resonance imaging (MRI) of amyloid. Furthermore, the pathological assessment was conducted on the same brain slices to evaluate the correlation between the CEST MRI signal abnormality and ß-amyloid deposition detected by immunohistochemical staining. In our results, the Z-spectra revealed an apparent CEST effect that peaked at approximately 6 ppm. APP/PS1 mice as young as 9 months injected with PiB showed a significantly higher CEST effect compared to the control groups. The hyperintense region was correlated with the Aß deposition shown by pathological staining. In conclusion, repurposing the PET probe prototype for CEST MRI imaging is feasible and enables label- and radiation-free detection of the amyloid while maintaining the sensitivity and specificity of the ligand. This study opens the door to developing CEST probes based on PET probe prototypes.

Effects of MRI scanner manufacturers in classification tasks with deep learning models.

Deep learning has become a leading subset of machine learning and has been successfully employed in diverse areas, ranging from natural language processing to medical image analysis. In medical imaging, researchers have progressively turned towards multi-center neuroimaging studies to address complex questions in neuroscience, leveraging larger sample sizes and aiming to enhance the accuracy of deep learning models. However, variations in image pixel/voxel characteristics can arise between centers due to factors including differences in magnetic resonance imaging scanners. Such variations create challenges, particularly inconsistent performance in machine learning-based approaches, often referred to as domain shift, where the trained models fail to achieve satisfactory or improved results when confronted with dissimilar test data. This study analyzes the performance of multiple disease classification tasks using multi-center MRI data obtained from three widely used scanner manufacturers (GE, Philips, and Siemens) across several deep learning-based networks. Furthermore, we investigate the efficacy of mitigating scanner vendor effects using ComBat-based harmonization techniques when applied to multi-center datasets of 3D structural MR images. Our experimental results reveal a substantial decline in classification performance when models trained on one type of scanner manufacturer are tested with data from different manufacturers. Moreover, despite applying ComBat-based harmonization, the harmonized images do not demonstrate any noticeable performance enhancement for disease classification tasks.

DSMRI: Domain Shift Analyzer for Multi-Center MRI Datasets.

In medical research and clinical applications, the utilization of MRI datasets from multiple centers has become increasingly prevalent. However, inherent variability between these centers presents challenges due to domain shift, which can impact the quality and reliability of the analysis. Regrettably, the absence of adequate tools for domain shift analysis hinders the development and validation of domain adaptation and harmonization techniques. To address this issue, this paper presents a novel Domain Shift analyzer for MRI (DSMRI) framework designed explicitly for domain shift analysis in multi-center MRI datasets. The proposed model assesses the degree of domain shift within an MRI dataset by leveraging various MRI-quality-related metrics derived from the spatial domain. DSMRI also incorporates features from the frequency domain to capture low- and high-frequency information about the image. It further includes the wavelet domain features by effectively measuring the sparsity and energy present in the wavelet coefficients. Furthermore, DSMRI introduces several texture features, thereby enhancing the robustness of the domain shift analysis process. The proposed framework includes visualization techniques such as t-SNE and UMAP to demonstrate that similar data are grouped closely while dissimilar data are in separate clusters. Additionally, quantitative analysis is used to measure the domain shift distance, domain classification accuracy, and the ranking of significant features. The effectiveness of the proposed approach is demonstrated using experimental evaluations on seven large-scale multi-site neuroimaging datasets.

SF2Former: Amyotrophic lateral sclerosis identification from multi-center MRI data using spatial and frequency fusion transformer.

Amyotrophic Lateral Sclerosis (ALS) is a complex neurodegenerative disorder characterized by motor neuron degeneration. Significant research has begun to establish brain magnetic resonance imaging (MRI) as a potential biomarker to diagnose and monitor the state of the disease. Deep learning has emerged as a prominent class of machine learning algorithms in computer vision and has shown successful applications in various medical image analysis tasks. However, deep learning methods applied to neuroimaging have not achieved superior performance in classifying ALS patients from healthy controls due to insignificant structural changes correlated with pathological features. Thus, a critical challenge in deep models is to identify discriminative features from limited training data. To address this challenge, this study introduces a framework called SF2Former, which leverages the power of the vision transformer architecture to distinguish ALS subjects from the control group by exploiting the long-range relationships among image features. Additionally, spatial and frequency domain information is combined to enhance the network's performance, as MRI scans are initially captured in the frequency domain and then converted to the spatial domain. The proposed framework is trained using a series of consecutive coronal slices and utilizes pre-trained weights from ImageNet through transfer learning. Finally, a majority voting scheme is employed on the coronal slices of each subject to generate the final classification decision. The proposed architecture is extensively evaluated with multi-modal neuroimaging data (i.e., T1-weighted, R2*, FLAIR) using two well-organized versions of the Canadian ALS Neuroimaging Consortium (CALSNIC) multi-center datasets. The experimental results demonstrate the superiority of the proposed strategy in terms of classification accuracy compared to several popular deep learning-based techniques.

Signal-to-noise ratio penalties from a loss of stimulated echoes when using slab-selective excitation in three-dimensional fast spin echo imaging with long echo trains.

Three-dimensional fast spin echo imaging with long echo trains combines high resolution with reasonable acquisition times and reduced specific absorption rate due to low refocusing flip angles. Typically, an entire volume is encoded (nonselective excitation) or localization can be performed with slab select excitation, which uses a long 90° pulse for precise localization, followed by a preliminary nonselective 180° pulse bounded by spoiler gradients to destroy signal outside of the volume of interest. Subsequent flip angles in the train are nonselective and identical between the two methods. The inclusion of the initial selective pulse and spoiler gradients results in a signal-to-noise ratio (SNR) penalty for slab selection, beyond the slice-averaging dependence, arising from a loss of stimulated echoes. SNR differences are explored using Bloch equation simulations of a T2-weighted 96 echo train sequence with varying parameters including T2, T1, and B1+ and compared with phantom and in vivo brain, neck, and knee experiments. In vivo SNR measurements in the three regions showed a maximum decrease of selective SNR by 29% (gastrocnemius muscle), 25% (pons), and 22% (globus pallidus), despite similar experimental parameters to nonselective experiments. Decreased SNR was compounded by B1+ variation affecting prescribed flip angles with further smaller reductions with T2 and T1 times. In conclusion, the elimination of coherences via the preliminary nominal 180° pulse and spoiler gradients in addition to the extended echo timing from the long excitation pulse resulted in a reduction in SNR compared with the nonselective case. Consideration of the required SNR and chosen anatomy as well as sequence restrictions should be weighed before choosing slab-selective excitation.

Longitudinal hippocampal diffusion-weighted imaging and T2 relaxometry demonstrate regional abnormalities which are stable and predict subfield pathology in temporal lobe epilepsy.

OBJECTIVE: High-resolution (1 mm isotropic) diffusion tensor imaging (DTI) of the hippocampus in temporal lobe epilepsy (TLE) patients has shown patterns of hippocampal subfield diffusion abnormalities, which were consistent with hippocampal sclerosis (HS) subtype on surgical histology. The objectives of this longitudinal imaging study were to determine the stability of focal hippocampus diffusion changes over time in TLE patients, compare diffusion and quantitative T2 abnormalities of the sclerotic hippocampus, and correlate presurgical mean diffusivity (MD) and T2 maps with postsurgical histology. METHODS: Nineteen TLE patients and 19 controls underwent two high-resolution (1 mm isotropic) DTI and 1.1 × 1.1 × 1 mm3  T2 relaxometry scans (in a subset of 16 TLE patients and 9 controls) of the hippocampus at 3T, with a 2.6 ± 0.8 year inter-scan interval. Within-participant hippocampal volume, MD and T2 were compared between the scans. Contralateral hippocampal changes 2.3 ± 1.0 years after surgery and ipsilateral preoperative MD maps versus postoperative subfield histopathology were evaluated in eight patients who underwent surgical resection of the hippocampus. RESULTS: Reduced volume and elevated MD and T2 of sclerotic hippocampi remained unchanged between longitudinal scans. Focal regions of elevated MD and T2 in bilateral hippocampi of HS TLE were detected consistently at both scans. Regions of high MD and T2 correlated and remained consistent over time. Volume, MD, and T2 remained unchanged in postoperative contralateral hippocampus. Regional elevations of MD identified subfield neuron loss on postsurgical histology with 88% sensitivity and 88% specificity. Focal T2 elevations identified subfield neuron loss with 75% sensitivity and 88% specificity. SIGNIFICANCE: Diffusion and T2 abnormalities in ipsilateral and contralateral hippocampi remained unchanged between the scans suggesting permanent microstructural alterations. MD and T2 demonstrated good sensitivity and specificity to detect hippocampal subfield neuron loss on postsurgical histology, supporting the potential that high-resolution hippocampal DTI and T2 could be used to diagnose HS subtype before surgery.

Quantitative Susceptibility Mapping Changes Relate to Gait Issues in Parkinson's Disease.

BACKGROUND: Quantitative susceptibility mapping (QSM) demonstrates elevated iron content in Parkinson's disease (PD) patients within the basal ganglia, though it has infrequently been studied in relation to gait difficulties including freezing of gait (FOG). Our purpose was to relate QSM of basal ganglia and extra-basal ganglia structures with qualitative and quantitative gait measures in PD. METHODS: This case-control study included PD and cognitively unimpaired (CU) participants from the Comprehensive Assessment of Neurodegeneration and Dementia study. Whole brain QSM was acquired at 3T. Region of interests (ROIs) were drawn blinded manually in the caudate nucleus, putamen, globus pallidus, pulvinar nucleus of the thalamus, red nucleus, substantia nigra, and dentate nucleus. Susceptibilities of ROIs were compared between PD and CU. Items from the FOG questionnaire and quantitative gait measures from PD participants were compared to susceptibilities. RESULTS: Twenty-nine participants with PD and 27 CU participants were included. There was no difference in susceptibility values in any ROI when comparing CU versus PD (p > 0.05 for all). PD participants with gait impairment (n = 23) had significantly higher susceptibility in the putamen (p = 0.008), red nucleus (p = 0.01), and caudate nucleus (p = 0.03) compared to those without gait impairment (n = 6). PD participants with FOG (n = 12) had significantly higher susceptibility in the globus pallidus (p = 0.03) compared to those without FOG (n = 17). Among quantitative gait measures, only stride time variability was significantly different between those with and without FOG (p = 0.04). CONCLUSION: Susceptibilities in basal ganglia and extra-basal ganglia structures are related to qualitative measures of gait impairment and FOG in PD.

Inline dual-echo T2 quantification in brain using a fast mapping reconstruction technique.

T2 mapping from 2D proton density and T2-weighted images (PD-T2) using Bloch equation simulations can be time consuming and introduces a latency between image acquisition and T2 map production. A fast T2 mapping reconstruction method is investigated and compared with a previous modeling approach to reduce computation time and allow inline T2 maps on the MRI console. Brain PD-T2 images from five multiple sclerosis patients were used to compare T2 map reconstruction times between the new subtraction method and the Euclidean norm minimization technique. Bloch equation simulations were used to create the lookup table for decay curve matching in both cases. Agreement of the two techniques used Bland-Altman analysis for investigating individual subsets of data and all image points in the five volumes (meta-analysis). The subtraction method resulted in an average reduction of computation time for single slices from 134 s (minimization method) to 0.44 s. Comparing T2 values between the subtraction and minimization methods resulted in a confidence interval ranging from -0.06 to 0.06 ms (95% of values were within ± 0.06 ms between the techniques). Using identical reconstruction code based on the subtraction method, inline T2 maps were produced from PD-T2 images directly on the scanner console. The excellent agreement between the two methods permits the subtraction technique to be interchanged with the previous method, reducing computation time and allowing inline T2 map reconstruction based on Bloch simulations directly on the scanner.

Automated detection of intracranial artery stenosis and occlusion in magnetic resonance angiography: A preliminary study based on deep learning.

BACKGROUND AND OBJECTIVES: Intracranial atherosclerotic stenosis of a major intracranial artery is the common cause of ischemic stroke. We evaluate the feasibility of using deep learning to automatically detect intracranial arterial steno-occlusive lesions from time-of-flight magnetic resonance angiography. METHODS: In a retrospective study, magnetic resonance images with radiological reports of intracranial arterial stenosis and occlusion were extracted. The images were randomly divided into a training set and a test set. The manual annotation of lesions with a bounding box labeled "moderate stenosis," "severe stenosis," "occlusion," and "absence of signal" was considered as ground truth. A deep learning algorithm based on you only look once version 5 (YOLOv5) detection model was developed with the training set, and its sensitivity and positive predictive values to detect lesions were evaluated in the test set. RESULTS: A dataset of 200 examinations consisted of a total of 411 lesions-242 moderate stenoses, 84 severe stenoses, 70 occlusions, and 15 absence of signal. The magnetic resonance images contained 291 lesions in the training set and 120 lesions in the test set. The sensitivity and positive predictive values were 64.2 and 83.7%, respectively. The detection sensitivity in relation to the location was greatest in the internal carotid artery (86.2%). CONCLUSIONS: Applying deep learning algorithms in the automated detection of intracranial arterial steno-occlusive lesions from time-of-flight magnetic resonance angiography is feasible and has great potential.

Quantitative susceptibility-weighted imaging in presence of strong susceptibility sources: Application to hemorrhage.

PURPOSE: To optimize quantitative susceptibility-weighted imaging also known as true susceptibility-weighted imaging (tSWI) for strong susceptibility sources like hemorrhage and compare to standard susceptibility-weighted imaging (SWI) and quantitative susceptibility mapping (QSM). METHODS: Ten patients with known intracerebral hemorrhage (ICH) were scanned using a 3D SWI sequence. The magnitude and phase images were utilized to compute QSM, tSWI and SWI images. tSWI parameters including the upper threshold for creating susceptibility-weighted masks and the multiplication factor were optimized for hemorrhage depiction. Combined tSWI was also computed with independent optimized parameters for both veins and hemorrhagic regions. tSWI results were compared to SWI and QSM utilizing region-of-interest measurements, Pearson's correlation and Kruskal-Wallis test. RESULTS: Fifteen hemorrhages were found, with mean susceptibility 0.81 ± 0.37 ppm. Unlike SWI which utilizes a phase mask, tSWI uses a mask computed from QSM. In tSWI, the weighted mask required an extended upper threshold far beyond the standard level for more effective visualization of hemorrhage texture. The upper threshold was set to the mean maximum susceptibility in the hemorrhagic region (3.24 ppm) with a multiplication factor of 2. The blooming effect, seen in SWI, was observed to be larger in hemorrhages with higher susceptibility values (r = 0.78, p < 0.001) with reduced blooming on tSWI. On SWI, 4 out of 15 hemorrhages showed phase wrap artifacts in the hemorrhagic region and all patients showed some phase wraps in the air-tissue interface near the auditory and frontal sinuses. These phase wrap artifacts were absent on tSWI. In hemorrhagic regions, a higher correlation was observed between the actual susceptibility values and mean gray value for tSWI (r = -0.93, p < 0.001) than SWI (r = -0.87, p < 0.001). CONCLUSION: In hemorrhage, tSWI minimizes both blooming effects and phase wrap artifacts observed in SWI. However, unlike SWI, tSWI requires an altered upper threshold for best hemorrhage depiction that greatly differs from the standard value. tSWI can be used as a complementary technique for visualizing hemorrhage along with SWI.

Multisite reproducibility of quantitative susceptibility mapping and effective transverse relaxation rate in deep gray matter at 3 T using locally optimized sequences in 24 traveling heads.

Iron concentration in the human brain plays a crucial role in several neurodegenerative diseases and can be monitored noninvasively using quantitative susceptibility mapping (QSM) and effective transverse relaxation rate (R2 *) mapping from multiecho T2 *-weighted images. Large population studies enable better understanding of pathologies and can benefit from pooling multisite data. However, reproducibility may be compromised between sites and studies using different hardware and sequence protocols. This work investigates QSM and R2 * reproducibility at 3 T using locally optimized sequences from three centers and two vendors, and investigates possible reduction of cross-site variability through postprocessing approaches. Twenty-four healthy subjects traveled between three sites and were scanned twice at each site. Scan-rescan measurements from seven deep gray matter regions were used for assessing within-site and cross-site reproducibility using intraclass correlation coefficient (ICC) and within-subject standard deviation (SDw) measures. In addition, multiple QSM and R2 * postprocessing options were investigated with the aim to minimize cross-site sequence-related variations, including: mask generation approach, echo-timing selection, harmonizing spatial resolution, field map estimation, susceptibility inversion method, and linear field correction for magnitude images. The same-subject cross-site region of interest measurements for QSM and R2 * were highly correlated (R2 ≥ 0.94) and reproducible (mean ICC of 0.89 and 0.82 for QSM and R2 *, respectively). The mean cross-site SDw was 4.16 parts per billion (ppb) for QSM and 1.27 s-1 for R2 *. For within-site measurements of QSM and R2 *, the mean ICC was 0.97 and 0.87 and mean SDw was 2.36 ppb and 0.97 s-1 , respectively. The precision level is regionally dependent and is reduced in the frontal lobe, near brain edges, and in white matter regions. Cross-site QSM variability (mean SDw) was reduced up to 46% through postprocessing approaches, such as masking out less reliable regions, matching available echo timings and spatial resolution, avoiding the use of the nonconsistent magnitude contrast between scans in field estimation, and minimizing streaking artifacts.

Myelin water fraction mapping from multiple echo spin echoes and an independent B1+ map.

PURPOSE: Myelin water fraction (MWF) is often obtained from a multiple echo spin echo (MESE) sequence using multi-component T2 fitting with non-negative least squares. This process fits many unknowns including B1+ to produce a T2 spectrum for each voxel. Presented is an alternative using a rapid B1+ mapping sequence to supply B1+ for the MWF fitting procedure. METHODS: Effects of B1+ errors on MWF calculations were modeled for 2D and 3D MESE using Bloch and extended phase graph simulations, respectively. Variations in SNR and relative refocusing widths were tested. Human brain experiments at 3 T used 2D MESE and an independent B1+ map. MWF maps were produced with the standard approach and with the use of the independent B1+ map. Differences in B1+ and mean MWF in specific brain regions were compared. RESULTS: For 2D MESE, MWF with the standard method was strongly affected by B1+ misestimations arising from limited SNR and response asymmetry around 180°, which decreased with increasing relative refocusing width. Using an independent B1+ map increased mean MWF and decreased coefficient of variation. Notable differences in vivo in 2D MESE were in areas of high B1+ such as thalamus and splenium where mean MWF increased by 88% and 31%, respectively (P < 0.001). Simulations also demonstrated the advantages of this approach for 3D MESE when SNR is <500. CONCLUSION: For 2D MESE, because of increased complexity of decay curves and limited SNR, supplying B1+ improves MWF results in peripheral and central brain regions where flip angles differ substantially from 180°.

Characterization of B1+ field variation in brain at 3 T using 385 healthy individuals across the lifespan.

PURPOSE: The transmit field B1+ at 3 T in brain affects the spatial uniformity and contrast of most image acquisitions. Here, B1+ spatial variation in brain at 3 T is characterized in a large healthy population. METHODS: Bloch-Siegert B1+ maps were acquired at 3 T from 385 healthy subjects aged 5-90 years on a single MRI system. After transforming all B1+ maps to a standard brain atlas space, region-of-interest analysis was performed, and intersubject voxel-wise coefficient of variation was calculated across the whole brain. The B1+ variability due to age and brain size was studied separately in males and females, along with B1+ variability due to nonideal transmit calibration. RESULTS: The voxel-based mean coefficient of variation was 4.0% across all subjects, and the difference in B1+ between central (left thalamus) and outer regions (left frontal gray matter) was 24.2% ± 2.3%. The least intersubject variability occurred in central regions, whereas regions toward brain edges increased markedly in variation. The B1+ variability with age was mostly attributed to lifespan changes in CSF volume (which alters brain conductivity) and head orientation. Larger brain size correlated with more B1+ inhomogeneity (p < .001). Varying head position and anatomy resulted in an inaccurate transmit calibration. CONCLUSION: In standard atlas space, intersubject B1+ variability at 3 T was relatively small in a large population aged 5-90 years. The B1+ varied with age-related changes of CSF volume and head orientation, as well as differences in brain size and transmit calibration.

T2 quantification in brain using 3D fast spin-echo imaging with long echo trains.

PURPOSE: Three-dimensional fast spin-echo (FSE) sequences commonly use very long echo trains (>64 echoes) and severely reduced refocusing angles. They are increasingly used in brain exams due to high, isotropic resolution and reasonable scan time when using long trains and short interecho spacing. In this study, T2 quantification in 3D FSE is investigated to achieve increased resolution when comparing with established 2D (proton-density dual-echo and multi-echo spin-echo) methods. METHODS: The FSE sequence design was explored to use long echo trains while minimizing T2 fitting error and maintaining typical proton density and T2 -weighted contrasts. Constant and variable flip angle trains were investigated using extended phase graph and Bloch equation simulations. Optimized parameters were analyzed in phantom experiments and validated in vivo in comparison to 2D methods for eight regions of interest in brain, including deep gray-matter structures and white-matter tracts. RESULTS: Phantom and healthy in vivo brain T2 measurements showed that optimized variable echo-train 3D FSE performs similarly to previous 2D methods, while achieving three-fold-higher slice resolution, evident visually in the 3D T2 maps. Optimization resulted in better T2 fitting and compared well with standard multi-echo spin echo (within the 8-ms confidence limits defined based on Bland-Altman analysis). CONCLUSION: T2 mapping using 3D FSE with long echo trains and variable refocusing angles provides T2 accuracy in agreement with 2D methods with additional high-resolution benefits, allowing isotropic views while avoiding incidental magnetization transfer effects. Consequently, optimized 3D sequences should be considered when choosing T2 mapping methods for high anatomic detail.

R2* and quantitative susceptibility mapping in deep gray matter of 498 healthy controls from 5 to 90 years.

Putative MRI markers of iron in deep gray matter have demonstrated age related changes during discrete periods of healthy childhood or adulthood, but few studies have included subjects across the lifespan. This study reports both transverse relaxation rate (R2*) and quantitative susceptibility mapping (QSM) of four primary deep gray matter regions (thalamus, putamen, caudate, and globus pallidus) in 498 healthy individuals aged 5-90 years. In the caudate, putamen, and globus pallidus, increases of QSM and R2* were steepest during childhood continuing gradually throughout adulthood, except caudate susceptibility which reached a plateau in the late 30s. The thalamus had a unique profile with steeper changes of R2* (reflecting additive effects of myelin and iron) than QSM during childhood, both reaching a plateau in the mid-30s to early 40s and decreasing thereafter. There were no hemispheric or sex differences for any region. Notably, both R2* and QSM values showed more inter-subject variability with increasing age from 5 to 90 years, potentially reflecting a common starting point in iron/myelination during childhood that diverges as a result of lifestyle and genetic factors that accumulate with age.

Bloch modelling enables robust T2 mapping using retrospective proton density and T2-weighted images from different vendors and sites.

T2 quantification is commonly attempted by applying an exponential fit to proton density (PD) and transverse relaxation (T2)-weighted fast spin echo (FSE) images. However, inter-site studies have noted systematic differences between vendors in T2 maps computed via standard exponential fitting due to imperfect slice refocusing, different refocusing angles and transmit field (B1+) inhomogeneity. We examine T2 mapping at 3T across 13 sites and two vendors in healthy volunteers from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database using both a standard exponential and a Bloch modelling approach. The standard exponential approach resulted in highly variable T2 values across different sites and vendors. The two-echo fitting method based on Bloch equation modelling of the pulse sequence with prior knowledge of the nominal refocusing angles, slice profiles, and estimated B1+ maps yielded similar T2 values across sites and vendors by accounting for the effects of indirect and stimulated echoes. By modelling the actual refocusing angles used, T2 quantification from PD and T2-weighted images can be applied in studies across multiple sites and vendors.