Repeat it without me: Crowdsourcing the T1 mapping common ground via the ISMRM reproducibility challenge.

PURPOSE: T1 mapping is a widely used quantitative MRI technique, but its tissue-specific values remain inconsistent across protocols, sites, and vendors. The ISMRM Reproducible Research and Quantitative MR study groups jointly launched a challenge to assess the reproducibility of a well-established inversion-recovery T1 mapping technique, using acquisition details from a seminal T1 mapping paper on a standardized phantom and in human brains. METHODS: The challenge used the acquisition protocol from Barral et al. (2010). Researchers collected T1 mapping data on the ISMRM/NIST phantom and/or in human brains. Data submission, pipeline development, and analysis were conducted using open-source platforms. Intersubmission and intrasubmission comparisons were performed. RESULTS: Eighteen submissions (39 phantom and 56 human datasets) on scanners by three MRI vendors were collected at 3 T (except one, at 0.35 T). The mean coefficient of variation was 6.1% for intersubmission phantom measurements, and 2.9% for intrasubmission measurements. For humans, the intersubmission/intrasubmission coefficient of variation was 5.9/3.2% in the genu and 16/6.9% in the cortex. An interactive dashboard for data visualization was also developed: https://rrsg2020.dashboards.neurolibre.org. CONCLUSION: The T1 intersubmission variability was twice as high as the intrasubmission variability in both phantoms and human brains, indicating that the acquisition details in the original paper were insufficient to reproduce a quantitative MRI protocol. This study reports the inherent uncertainty in T1 measures across independent research groups, bringing us one step closer to a practical clinical baseline of T1 variations in vivo.

Mitochondrial Creatine Kinase Attenuates Pathologic Remodeling in Heart Failure.

BACKGROUND: Abnormalities in cardiac energy metabolism occur in heart failure (HF) and contribute to contractile dysfunction, but their role, if any, in HF-related pathologic remodeling is much less established. CK (creatine kinase), the primary muscle energy reserve reaction which rapidly provides ATP at the myofibrils and regenerates mitochondrial ADP, is down-regulated in experimental and human HF. We tested the hypotheses that pathologic remodeling in human HF is related to impaired cardiac CK energy metabolism and that rescuing CK attenuates maladaptive hypertrophy in experimental HF. METHODS: First, in 27 HF patients and 14 healthy subjects, we measured cardiac energetics and left ventricular remodeling using noninvasive magnetic resonance 31P spectroscopy and magnetic resonance imaging, respectively. Second, we tested the impact of metabolic rescue with cardiac-specific overexpression of either Ckmyofib (myofibrillar CK) or Ckmito (mitochondrial CK) on HF-related maladaptive hypertrophy in mice. RESULTS: In people, pathologic left ventricular hypertrophy and dilatation correlate closely with reduced myocardial ATP levels and rates of ATP synthesis through CK. In mice, transverse aortic constriction-induced left ventricular hypertrophy and dilatation are attenuated by overexpression of CKmito, but not by overexpression of CKmyofib. CKmito overexpression also attenuates hypertrophy after chronic isoproterenol stimulation. CKmito lowers mitochondrial reactive oxygen species, tissue reactive oxygen species levels, and upregulates antioxidants and their promoters. When the CK capacity of CKmito-overexpressing mice is limited by creatine substrate depletion, the protection against pathologic remodeling is lost, suggesting the ADP regenerating capacity of the CKmito reaction rather than CK protein per se is critical in limiting adverse HF remodeling. CONCLUSIONS: In the failing human heart, pathologic hypertrophy and adverse remodeling are closely related to deficits in ATP levels and in the CK energy reserve reaction. CKmito, sitting at the intersection of cardiac energetics and redox balance, plays a crucial role in attenuating pathologic remodeling in HF. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT00181259.

Cerebrovascular Effects of Lower Body Negative Pressure at 3T MRI: Implications for Long-Duration Space Travel.

BACKGROUND: Optic disc edema develops in most astronauts during long-duration spaceflight. It is hypothesized to result from weightlessness-induced venous congestion of the head and neck and is an unresolved health risk of space travel. PURPOSE: Determine if short-term application of lower body negative pressure (LBNP) could reduce internal jugular vein (IJV) expansion associated with the supine posture without negatively impacting cerebral perfusion or causing IJV flow stasis. STUDY TYPE: Prospective. SUBJECTS: Nine healthy volunteers (six women). FIELD STRENGTH/SEQUENCE: 3T/cine two-dimensional phase-contrast gradient echo; pseudo-continuous arterial spin labeling single-shot gradient echo echo-planar. ASSESSMENT: The study was performed with two sequential conditions in randomized order: supine posture and supine posture with 25 mmHg LBNP (LBNP25 ). LBNP was achieved by enclosing the lower extremities in a semi-airtight acrylic chamber connected to a vacuum. Heart rate, bulk cerebrovasculature flow, IJV cross-sectional area, fractional IJV outflow relative to arterial inflow, and cerebral perfusion were assessed in each condition. STATISTICAL TESTS: Paired t-tests were used to compare measurement means across conditions. Significance was defined as P < 0.05. RESULTS: LBNP25 significantly increased heart rate from 64 ± 9 to 71 ± 8 beats per minute and significantly decreased IJV cross-sectional area, IJV outflow fraction, cerebral arterial flow rate, and cerebral arterial stroke volume from 1.28 ± 0.64 to 0.56 ± 0.31 cm2 , 0.75 ± 0.20 to 0.66 ± 0.28, 780 ± 154 to 708 ± 137 mL/min and 12.2 ± 2.8 to 9.7 ± 1.7 mL/cycle, respectively. During LBNP25 , there was no significant change in gray or white matter cerebral perfusion (P = 0.26 and P = 0.24 respectively) and IJV absolute mean peak flow velocity remained ≥4 cm/sec in all subjects. DATA CONCLUSION: Short-term application of LBNP25 reduced IJV expansion without decreasing cerebral perfusion or inducing IJV flow stasis. LEVEL OF EVIDENCE: 1 TECHNICAL EFFICACY STAGE: 1.

Deep learning segmentation of gadolinium-enhancing lesions in multiple sclerosis.

OBJECTIVE: The aim of this study is to assess the performance of deep learning convolutional neural networks (CNNs) in segmenting gadolinium-enhancing lesions using a large cohort of multiple sclerosis (MS) patients. METHODS: A three-dimensional (3D) CNN model was trained for segmentation of gadolinium-enhancing lesions using multispectral magnetic resonance imaging data (MRI) from 1006 relapsing-remitting MS patients. The network performance was evaluated for three combinations of multispectral MRI used as input: (U5) fluid-attenuated inversion recovery (FLAIR), T2-weighted, proton density-weighted, and pre- and post-contrast T1-weighted images; (U2) pre- and post-contrast T1-weighted images; and (U1) only post-contrast T1-weighted images. Segmentation performance was evaluated using the Dice similarity coefficient (DSC) and lesion-wise true-positive (TPR) and false-positive (FPR) rates. Performance was also evaluated as a function of enhancing lesion volume. RESULTS: The DSC/TPR/FPR values averaged over all the enhancing lesion sizes were 0.77/0.90/0.23 using the U5 model. These values for the largest enhancement volumes (>500 mm3) were 0.81/0.97/0.04. For U2, the average DSC/TPR/FPR values were 0.72/0.86/0.31. Comparable performance was observed with U1. For all types of input, the network performance degraded with decreased enhancement size. CONCLUSION: Excellent segmentation of enhancing lesions was observed for enhancement volume ⩾70 mm3. The best performance was achieved when the input included all five multispectral image sets.

Deep Learning for Predicting Enhancing Lesions in Multiple Sclerosis from Noncontrast MRI.

Background Enhancing lesions on MRI scans obtained after contrast material administration are commonly thought to represent disease activity in multiple sclerosis (MS); it is desirable to develop methods that can predict enhancing lesions without the use of contrast material. Purpose To evaluate whether deep learning can predict enhancing lesions on MRI scans obtained without the use of contrast material. Materials and Methods This study involved prospective analysis of existing MRI data. A convolutional neural network was used for classification of enhancing lesions on unenhanced MRI scans. This classification was performed for each slice, and the slice scores were combined by using a fully connected network to produce participant-wise predictions. The network input consisted of 1970 multiparametric MRI scans from 1008 patients recruited from 2005 to 2009. Enhanced lesions on postcontrast T1-weighted images served as the ground truth. The network performance was assessed by using fivefold cross-validation. Statistical analysis of the network performance included calculation of lesion detection rates and areas under the receiver operating characteristic curve (AUCs). Results MRI scans from 1008 participants (mean age, 37.7 years ± 9.7; 730 women) were analyzed. At least one enhancing lesion was observed in 519 participants. The sensitivity and specificity averaged across the five test sets were 78% ± 4.3 and 73% ± 2.7, respectively, for slice-wise prediction. The corresponding participant-wise values were 72% ± 9.0 and 70% ± 6.3. The diagnostic performances (AUCs) were 0.82 ± 0.02 and 0.75 ± 0.03 for slice-wise and participant-wise enhancement prediction, respectively. Conclusion Deep learning used with conventional MRI identified enhanced lesions in multiple sclerosis from images from unenhanced multiparametric MRI with moderate to high accuracy. © RSNA, 2019.

Deep-Learning-Based Neural Tissue Segmentation of MRI in Multiple Sclerosis: Effect of Training Set Size.

BACKGROUND: The dependence of deep-learning (DL)-based segmentation accuracy of brain MRI on the training size is not known. PURPOSE: To determine the required training size for a desired accuracy in brain MRI segmentation in multiple sclerosis (MS) using DL. STUDY TYPE: Retrospective analysis of MRI data acquired as part of a multicenter clinical trial. STUDY POPULATION: In all, 1008 patients with clinically definite MS. FIELD STRENGTH/SEQUENCE: MRIs were acquired at 1.5T and 3T scanners manufactured by GE, Philips, and Siemens with dual turbo spin echo, FLAIR, and T1 -weighted turbo spin echo sequences. ASSESSMENT: Segmentation results using an automated analysis pipeline and validated by two neuroimaging experts served as the ground truth. A DL model, based on a fully convolutional neural network, was trained separately using 16 different training sizes. The segmentation accuracy as a function of the training size was determined. These data were fitted to the learning curve for estimating the required training size for desired accuracy. STATISTICAL TESTS: The performance of the network was evaluated by calculating the Dice similarity coefficient (DSC), and lesion true-positive and false-positive rates. RESULTS: The DSC for lesions showed much stronger dependency on the sample size than gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). When the training size was increased from 10 to 800 the DSC values varied from 0.00 to 0.86 ± 0.016 for T2 lesions, 0.87 ± 009 to 0.94 ± 0.004 for GM, 0.86 ± 0.08 to 0.94 ± 0.005 for WM, and 0.91 ± 0.009 to 0.96 ± 0.003 for CSF. DATA CONCLUSION: Excellent segmentation was achieved with a training size as small as 10 image volumes for GM, WM, and CSF. In contrast, a training size of at least 50 image volumes was necessary for adequate lesion segmentation. LEVEL OF EVIDENCE: 1 Technical Efficacy Stage: 1 J. Magn. Reson. Imaging 2020;51:1487-1496.

Brain and lesion segmentation in multiple sclerosis using fully convolutional neural networks: A large-scale study.

OBJECTIVE: To investigate the performance of deep learning (DL) based on fully convolutional neural network (FCNN) in segmenting brain tissues in a large cohort of multiple sclerosis (MS) patients. METHODS: We developed a FCNN model to segment brain tissues, including T2-hyperintense MS lesions. The training, validation, and testing of FCNN were based on ~1000 magnetic resonance imaging (MRI) datasets acquired on relapsing-remitting MS patients, as a part of a phase 3 randomized clinical trial. Multimodal MRI data (dual-echo, FLAIR, and T1-weighted images) served as input to the network. Expert validated segmentation was used as the target for training the FCNN. We cross-validated our results using the leave-one-center-out approach. RESULTS: We observed a high average (95% confidence limits) Dice similarity coefficient for all the segmented tissues: 0.95 (0.92-0.98) for white matter, 0.96 (0.93-0.98) for gray matter, 0.99 (0.98-0.99) for cerebrospinal fluid, and 0.82 (0.63-1.0) for T2 lesions. High correlations between the DL segmented tissue volumes and ground truth were observed (R2 > 0.92 for all tissues). The cross validation showed consistent results across the centers for all tissues. CONCLUSION: The results from this large-scale study suggest that deep FCNN can automatically segment MS brain tissues, including lesions, with high accuracy.

Automated image quality evaluation of structural brain MRI using an ensemble of deep learning networks.

BACKGROUND: Deep learning (DL) is a promising methodology for automatic detection of abnormalities in brain MRI. PURPOSE: To automatically evaluate the quality of multicenter structural brain MRI images using an ensemble DL model based on deep convolutional neural networks (DCNNs). STUDY TYPE: Retrospective. POPULATION: The study included 1064 brain images of autism patients and healthy controls from the Autism Brain Imaging Data Exchange (ABIDE) database. MRI data from 110 multiple sclerosis patients from the CombiRx study were included for independent testing. SEQUENCE: T1 -weighted MR brain images acquired at 3T. ASSESSMENT: The ABIDE data were separated into training (60%), validation (20%), and testing (20%) sets. The ensemble DL model combined the results from three cascaded networks trained separately on the three MRI image planes (axial, coronal, and sagittal). Each cascaded network consists of a DCNN followed by a fully connected network. The quality of image slices from each plane was evaluated by the DCNN and the resultant image scores were combined into a volumewise quality rating using the fully connected network. The DL predicted ratings were compared with manual quality evaluation by two experts. STATISTICAL TESTS: Receiver operating characteristic (ROC) curve, area under ROC curve (AUC), sensitivity, specificity, accuracy, and positive (PPV) and negative (NPV) predictive values. RESULTS: The AUC, sensitivity, specificity, accuracy, PPV, and NPV for image quality evaluation of the ABIDE test set using the ensemble model were 0.90, 0.77, 0.85, 0.84, 0.42, and 0.96, respectively. On the CombiRx set the same model achieved performance of 0.71, 0.41, 0.84, 0.73, 0.48, and 0.80. DATA CONCLUSION: This study demonstrated the high accuracy of DL in evaluating image quality of structural brain MRI in multicenter studies. LEVEL OF EVIDENCE: 3 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;50:1260-1267.

Cardiac work is related to creatine kinase energy supply in human heart failure: a cardiovascular magnetic resonance spectroscopy study.

BACKGROUND: It has been hypothesized that the supply of chemical energy may be insufficient to fuel normal mechanical pump function in heart failure (HF). The creatine kinase (CK) reaction serves as the heart's primary energy reserve, and the supply of adenosine triphosphate (ATP flux) it provides is reduced in human HF. However, the relationship between the CK energy supply and the mechanical energy expended has never been quantified in the human heart. This study tests whether reduced CK energy supply is associated with reduced mechanical work in HF patients. METHODS: Cardiac mechanical work and CK flux in W/kg, and mechanical efficiency were measured noninvasively at rest using cardiac pressure-volume loops, magnetic resonance imaging and phosphorus spectroscopy in 14 healthy subjects and 27 patients with mild-to-moderate HF. RESULTS: In HF, the resting CK flux (126 ± 46 vs. 179 ± 50 W/kg, p < 0.002), the average (6.8 ± 3.1 vs. 10.1 ± 1.5 W/kg, p  <0.001) and the peak (32 ± 14 vs. 48 ± 8 W/kg, p < 0.001) cardiac mechanical work-rates, as well as the cardiac mechanical efficiency (53% ± 16 vs. 79% ± 3, p < 0.001), were all reduced by a third compared to healthy subjects. In addition, cardiac CK flux correlated with the resting peak and average mechanical power (p < 0.01), and with mechanical efficiency (p = 0.002). CONCLUSION: These first noninvasive findings showing that cardiac mechanical work and efficiency in mild-to-moderate human HF decrease proportionately with CK ATP energy supply, are consistent with the energy deprivation hypothesis of HF. CK energy supply exceeds mechanical work at rest but lies within a range that may be limiting with moderate activity, and thus presents a promising target for HF treatment. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT00181259 .

Patient-specific 3D FLAIR for enhanced visualization of brain white matter lesions in multiple sclerosis.

PURPOSE: To improve the conspicuity of white matter lesions (WMLs) in multiple sclerosis (MS) using patient-specific optimization of single-slab 3D fluid-attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI). MATERIALS AND METHODS: Sixteen MS patients were enrolled in a prospective 3.0T MRI study. FLAIR inversion time and echo time were automatically optimized for each patient during the same scan session based on measurements of the relative proton density and relaxation times of the brain tissues. The optimization criterion was to maximize the contrast between gray matter (GM) and white matter (WM), while suppressing cerebrospinal fluid. This criterion also helps increase the contrast between WMLs and WM. The performance of the patient-specific 3D FLAIR protocol relative to the fixed-parameter protocol was assessed both qualitatively and quantitatively. RESULTS: Patient-specific optimization achieved a statistically significant 41% increase in the GM-WM contrast ratio (P < 0.05) and 32% increase in the WML-WM contrast ratio (P < 0.01) compared with fixed-parameter FLAIR. The increase in WML-WM contrast ratio correlated strongly with echo time (P < 10-11 ). Two experienced neuroradiologists indicated substantially higher lesion conspicuity on the patient-specific FLAIR images over conventional FLAIR in 3-4 cases (intrarater correlation coefficient ICC = 0.72). In no case was the image quality of patient-specific FLAIR considered inferior to conventional FLAIR by any of the raters (ICC = 0.32). CONCLUSION: Changes in proton density and relaxation times render fixed-parameter FLAIR suboptimal in terms of lesion contrast. Patient-specific optimization of 3D FLAIR increases lesion conspicuity without scan time penalty, and has potential to enhance the detection of subtle and small lesions in MS. LEVEL OF EVIDENCE: 1 Technical Efficacy: Stage 1 J. MAGN. RESON. IMAGING 2017;46:557-564.

Automated patient-specific optimization of three-dimensional double-inversion recovery magnetic resonance imaging.

PURPOSE: To automatically optimize three-dimensional double-inversion recovery (3D-DIR) MRI of the brain on a patient-by-patient basis. METHODS: DIR is a powerful MRI technique that allows simultaneous suppression of white matter (WM) and cerebrospinal fluid (CSF) in brain imaging. Unfortunately, the tissue suppression is not always consistent across patients. We propose patient-specific optimization of WM suppression for improved gray matter (GM)-WM contrast. Relaxation times were measured in the same scan session, and through real time processing were used for calculating DIR inversion times for maximum tissue contrast. Signal evolution during the variable-flip-angle turbo-spin-echo readout was calculated using the extended phase graph algorithm. Patient-specific optimization was examined in five healthy volunteers and two multiple sclerosis patients. Two volunteers were scanned twice for reproducibility. The contrast ratios, GM signal-to-noise ratio (SNR), and image histogram were used to assess the performance of this patient-specific approach. RESULTS: Automated optimization of 3D-DIR was successfully completed in all experiments with processing time of ∼1 min. GM-WM contrast ratio tripled with the optimized DIR sequence, with only a 19% decrease in GM-CSF contrast and 30% SNR penalty. CONCLUSION: Patient-specific optimization is feasible and significantly improves GM-WM contrast on 3D-DIR with a moderate decrease in the GM SNR.

Optimal combination of FLAIR and T2-weighted MRI for improved lesion contrast in multiple sclerosis.

PURPOSE: Postacquisition combination of three-dimensional T2-weighted (T2w) and fluid-attenuated inversion recovery (FLAIR) images can improve the visualization of brain lesions in multiple sclerosis (MS). However, an optimal way to combine these images has not been described so far. The main objective of this study is to investigate an optimal combination of T2w and FLAIR to improve the conspicuity of MS lesions. MATERIALS AND METHODS: We determined the parameters for a generalized multiplicative image combination which maximize the contrast-to-noise ratio (CNR) between lesions and normal-appearing brain tissue through simulations and verified experimentally. MRI data from 11 MS patients acquired at 3 Tesla were retrospectively analyzed using the proposed approach and compared with conventional FLAIR, and to images obtained by direct multiplication of T2w and FLAIR (FLAIR2 ). Image quality was assessed by region-of-interest analysis. In addition, to evaluate the degree of cerebrospinal fluid (CSF) suppression, CSF-to-gray matter (CSF/GM) ratio was calculated. Reduction in global image contrast was assessed by computing the reduction in the contrast of mid-level intensity values. RESULTS: An optimal combination was found to be the third order expression: FLAIR3 = FLAIR1.55 × T2w1.45 . Compared with FLAIR, the lesion CNR was significantly increased by 1.9× (P < 0.005) and 2.5× (P < 0.001) using FLAIR2 and FLAIR3 , respectively. CSF/GM ratio was increased by 1.7× in FLAIR2 (P < 0.001) compared with FLAIR, while it was reduced to 0.7× on FLAIR3 (P < 0.05). The mid-intensity contrast was preserved on FLAIR2 (P = 0.2), and decreased by 29% on FLAIR3 (P < 0.001). CONCLUSION: These results show that the optimized combination of FLAIR and T2w can improve MS lesion conspicuity. J. Magn. Reson. Imaging 2016;44:1293-1300.

Two repetition time saturation transfer (TwiST) with spill-over correction to measure creatine kinase reaction rates in human hearts.

BACKGROUND: Phosphorus saturation transfer (ST) magnetic resonance spectroscopy can measure the rate of ATP generated from phosphocreatine (PCr) via creatine kinase (CK) in the human heart. Recently, the triple-repetition time ST (TRiST) method was introduced to measure the CK pseudo-first-order rate constant kf in three acquisitions. In TRiST, the longitudinal relaxation time of PCr while γ-ATP is saturated, T1`, is measured for each subject, but suffers from low SNR because the PCr signal is reduced due to exchange with saturated γ-ATP, and the short repetition time of one of the acquisitions. Here, a two-repetition time ST (TwiST) method is presented. In TwiST, the acquisition with γ-ATP saturation and short repetition time is dropped. Instead of measuring T1`, an intrinsic relaxation time T1 for PCr, T1 (intrinsic), is assumed. The objective was to validate TwiST measurements of CK kinetics in healthy subjects and patients with heart failure (HF). METHODS: Bloch equation simulations that included the effect of spillover irradiation on PCr were used to derive formulae for T1 (intrinsic) and kf measured by both TRiST and TwiST methods. Spillover was quantified from an unsaturated PCr measurement used in the current protocol for determining PCr and ATP concentrations. Cardiac TRiST and TwiST data were acquired at 3 T from 12 healthy and 17 HF patients. RESULTS: Simulations showed that both kf measured by TwiST and T1 (intrinsic) require spill-over corrections. In human heart at 3 T, the spill-over corrected T1 (intrinsic) = 8.4 ± 1.4 s (mean ± SD) independent of study group. TwiST and TRiST kf measurements were the same, but TwiST was 9 min faster. Spill-over corrected TwiST kf was 0.33 ± 0.08 s(-1) vs. 0.20 ± 0.06 s(-1) in healthy vs HF hearts, respectively (p < 0.0001). CONCLUSION: TwiST was validated against TRiST in the human heart at 3 T, generating the same results 9 min faster. TwiST detected significant reductions in CK kf in HF compared to healthy subjects, consistent with prior 1.5 T studies using different methodology.

Improving spectral quality in fetal brain magnetic resonance spectroscopy using constructive averaging.

PURPOSE: A common source of loss in signal-to-noise ratio (SNR) in fetal brain magnetic resonance spectroscopy (MRS) is from fetal movement and temporal magnetic field drift. We investigated the feasibility of using constructive averaging strategies for improving the spectral quality and recovering the SNR loss from these effects. MATERIALS AND METHODS: Eight fetuses, between 20 3/7 and 38 2/7 weeks' gestation, were scanned with MRS at 1.5 T. Single-voxel point-resolved spectroscopy of the fetal brain with TE = 144 ms (in one case additional TE = 288 ms) was performed in a dynamic mode, and individual spectra of 128 acquisitions were saved. With constructive averaging strategy individual acquisitions were corrected for phase variations and frequency drift before averaging. Constructively averaged spectra were compared to those using conventional averaging to evaluate differences in spectral quality and SNR. RESULTS: The definition of key metabolite peaks was qualitatively improved using constructive averaging, including the doublet structure of lactate in one case. Constructive averaging was associated with SNR increases, ranging from 11% to 40%, and the SNR further improved in one case when outliers from severe motion were rejected before averaging. CONCLUSION: Our results demonstrate the feasibility of using constructive averaging for improving SNR in fetal MRS, which is likely to improve the characterization of fetal brain metabolites.

Artificial Intelligence Assistive Software Tool for Automated Detection and Quantification of Amyloid-Related Imaging Abnormalities.

Importance: Amyloid-related imaging abnormalities (ARIA) are brain magnetic resonance imaging (MRI) findings associated with the use of amyloid-ß-directed monoclonal antibody therapies in Alzheimer disease (AD). ARIA monitoring is important to inform treatment dosing decisions and might be improved through assistive software. Objective: To assess the clinical performance of an artificial intelligence (AI)-based software tool for assisting radiological interpretation of brain MRI scans in patients monitored for ARIA. Design, Setting, and Participants: This diagnostic study used a multiple-reader multiple-case design to evaluate the diagnostic performance of radiologists assisted by the software vs unassisted. The study enrolled 16 US Board of Radiology-certified radiologists to perform radiological reading with (assisted) and without the software (unassisted). The study encompassed 199 retrospective cases, where each case consisted of a predosing baseline and a postdosing follow-up MRI of patients from aducanumab clinical trials PRIME, EMERGE, and ENGAGE. Statistical analysis was performed from April to July 2023. Exposures: Use of icobrain aria, an AI-based assistive software for ARIA detection and quantification. Main Outcomes and Measures: Coprimary end points were the difference in diagnostic accuracy between assisted and unassisted detection of ARIA-E (edema and/or sulcal effusion) and ARIA-H (microhemorrhage and/or superficial siderosis) independently, assessed with the area under the receiver operating characteristic curve (AUC). Results: Among the 199 participants included in this study of radiological reading performance, mean (SD) age was 70.4 (7.2) years; 105 (52.8%) were female; 23 (11.6%) were Asian, 1 (0.5%) was Black, 157 (78.9%) were White, and 18 (9.0%) were other or unreported race and ethnicity. Among the 16 radiological readers included, 2 were specialized neuroradiologists (12.5%), 11 were male individuals (68.8%), 7 were individuals working in academic hospitals (43.8%), and they had a mean (SD) of 9.5 (5.1) years of experience. Radiologists assisted by the software were significantly superior in detecting ARIA than unassisted radiologists, with a mean assisted AUC of 0.87 (95% CI, 0.84-0.91) for ARIA-E detection (AUC improvement of 0.05 [95% CI, 0.02-0.08]; P = .001]) and 0.83 (95% CI, 0.78-0.87) for ARIA-H detection (AUC improvement of 0.04 [95% CI, 0.02-0.07]; P = .001). Sensitivity was significantly higher in assisted reading compared with unassisted reading (87% vs 71% for ARIA-E detection; 79% vs 69% for ARIA-H detection), while specificity remained above 80% for the detection of both ARIA types. Conclusions and Relevance: This diagnostic study found that radiological reading performance for ARIA detection and diagnosis was significantly better when using the AI-based assistive software. Hence, the software has the potential to be a clinically important tool to improve safety monitoring and management of patients with AD treated with amyloid-ß-directed monoclonal antibody therapies.

Quantification of human high-energy phosphate metabolite concentrations at 3 T with partial volume and sensitivity corrections.

Practical noninvasive methods for the measurement of absolute metabolite concentrations are key to the assessment of the depletion of myocardial metabolite pools which occurs with several cardiac diseases, including infarction and heart failure. Localized MRS offers unique noninvasive access to many metabolites, but is often confounded by nonuniform sensitivity and partial volume effects in the large, poorly defined voxels commonly used for the detection of low-concentration metabolites with surface coils. These problems are exacerbated at higher magnetic field strengths by greater radiofrequency (RF) field inhomogeneity and differences in RF penetration with heteronuclear concentration referencing. An example is the (31)P measurement of cardiac adenosine triphosphate (ATP) and phosphocreatine (PCr) concentrations, which, although central to cardiac energetics, have not been measured at field strengths above 1.5 T. Here, practical acquisition and analysis protocols are presented for the quantification of [PCr] and [ATP] with one-dimensionally resolved surface coil spectra and concentration referencing at 3 T. The effects of nonuniform sensitivity and partial tissue volumes are addressed at 3 T by the application of MRI-based three-dimensional sensitivity weighting and tissue segmentation. The method is validated in phantoms of different sizes and concentrations, and used to measure [PCr] and [ATP] in healthy subjects. In calf muscle (n = 8), [PCr] = 24.7 ± 3.4 and [ATP] = 5.7 ± 1.3 µmol/g wet weight, whereas, in heart (n = 18), [PCr] = 10.4 ± 1.5 and [ATP] = 6.0 ± 1.1 µmol/g wet weight (all mean ± SD), consistent with previous reports at lower fields. The method enables, for the first time, the efficient, semi-automated quantification of high-energy phosphate metabolites in humans at 3 T with nonuniform excitation and detection.