BACKGROUND AND HYPOTHESIS:Hypertrophic cardiomyopathy (HCM) is considered a metabolic disease, but it is unknown whether all sarcomeric protein gene mutations lead to similar derangements of cardiac energy metabolism. Our prior studies showing differences in mitochondrial function in mouse models led us to hypothesize that HCM mutations lead to allele-specific metabolic pathway remodeling. In order to test this hypothesis, we examined cardiac substrate utilization using a triple tracer approach that simultaneously assays fatty acid (FA), ketone, and glucose oxidation, in R403Q-MyHC and R92W-TnT perfused hearts at established disease stage. METHODS:Mouse hearts (14-16 weeks old) were Langendorff-perfused to steady state 13C enrichment (∼30 min) while placed in a 14.1 T NMR magnet for in situ 31P spectroscopy. The perfusate contained substrates which each produce distinctively labeled acetyl-CoA: [1,6-13C2]glucose, [U13C]mixed fatty acids, and [1,3-13C2]β-hydroxybutyrate. At the end of perfusion, hearts were freeze clamped and extracted for 13C NMR isotopomer analysis. Oxygen concentration in afferent/efferent perfusate was measured using a Clark electrode, and used to compute myocardial oxygen consumption rate (MVO2). RESULTS:Using a model of citric acid cycle (CAC) metabolism that equates MVO2 with production of reducing equivalents, we observed significantly lower fatty acid (FA) oxidation and pyruvate dehydrogenase (PDH) flux in MyHC mutants, but similar substrate utilization in TnT mutants, when compared to respective controls. 31P-spectroscopy revealed significantly lower energy reserves reflected by lower PCr/ATP ratios in MyHC mutants, but similar PCr/ATP ratios in TnT mutants compared to controls. CONCLUSION:Cardiac metabolic flux measurements revealed an energy-deficient cardiac phenotype in R403Q-MyHC mutants, but not R92W-TnT mutants, reflecting allele-specific remodeling of cardiac energy metabolism. These results suggest that therapeutics aimed at downregulation of FA or carbohydrate metabolism may have mutation-specific effects.

Background/Objective: Longitudinal in vivo studies of murine xenograft models are widely utilized in oncology to study cancer biology and develop therapies. Magnetic resonance imaging (MRI) of these tumors is an invaluable tool for monitoring tumor growth and characterizing the tumors as well. Methods: In this work, a pipeline for automating the segmentation of xenografts in mouse models was developed. T2-weighted (T2-wt) MRI images from mice implanted with six different prostate cancer patient-derived xenografts (PDX) in the kidneys, liver, and tibia were used. The segmentation pipeline included a slice classifier to identify the slices that had tumors and subsequent training and validation using several U-Net-based segmentation architectures. Multiple combinations of the algorithm and training images for different sites were evaluated for inference quality. Results and Conclusions: The slice classifier network achieved 90% accuracy in identifying slices containing tumors. Among the various segmentation architectures tested, the dense residual recurrent U-Net achieved the highest performance in kidney tumors. When evaluated across the kidneys, tibia, and liver, this architecture performed the best when trained on all data as compared to training on only data from a single site (and inferring on a multi-site tumor images), achieving a Dice score of 0.924 across the test set.

PURPOSE:In this work, we adopt the MR fingerprinting (MRF) framework and leverage its flexibility in quantitative pulse sequence design to propose improved balanced steady-state free precession (bSSFP)-based hyperpolarized Carbon-13 (13C) acquisitions for robust metabolic conversion rate quantification. METHODS:Spectrally selective bSSFP-based acquisitions with variable RF excitation were implemented for [1-13C]pyruvate and used in conjunction with prior implementation of [1-13C]lactate selective bSSFP imaging. MRF framework parameter estimation was performed using dictionary-based template matching. Influences of bSSFP-based acquisitions and sigmoid RF excitation scheme were assessed with simulation experiments and Monte Carlo evaluation. Methods were then compared using experimental data from rat kidney acquired on a clinical 3 T scanner. RESULTS:Simulations indicated that combining bSSFP-based acquisitions and variable RF excitation (MRF-Sigmoid) exhibited bias <0.1% across the majority (86%) of combinations of pyruvate-to-lactate conversion rate (kPL) and noise level investigated when estimating kPL with the MRF framework. bSSFP-based experiments, with and without sigmoid excitation scheme, showed lower variance in fits at all levels of kPL and noise investigated compared to the method used in prior work by this group (hybrid gradient echo). Positive, linear correlations were found for in vivo voxel-wise estimates of kPL in healthy rat kidneys when comparing all experiment methods. MRF-Sigmoid experiment design increased pyruvate cumulative SNR by 3.5-fold over hybrid gradient echo while maintaining similar lactate cumulative SNR. CONCLUSION:The use of the MRF framework for kPL estimation demonstrates the feasibility of dictionary-based template matching and can be used to accurately estimate physiologically relevant kPL and improve cumulative SNR.

Magnetic resonance imaging of hyperpolarized (HP) [1-13C]pyruvate allows in-vivo assessment of metabolism and has translated into human studies across diseases at 15 centers worldwide. Consensus on best practice for multi-center studies is required to develop clinical applications. This paper presents the results of a 2-round formal consensus building exercise carried out by experts with HP [1-13C]pyruvate human study experience. Twenty-nine participants from 13 sites brought together expertise in pharmacy methods, MR physics, translational imaging, and data-analysis; with the goal of providing recommendations and best practice statements on conduct of multi-center human studies of HP [1-13C]pyruvate MRI. Overall, the group reached consensus on approximately two-thirds of 246 statements in the questionnaire, covering 'HP 13C-Pyruvate Preparation', 'MRI System Setup, Calibration, and Phantoms', 'Acquisition and Reconstruction', and 'Data Analysis and Quantification'. Consensus was present across categories, examples include that: (i) different HP pyruvate preparation methods could be used in human studies, but that the same release criteria have to be followed; (ii) site qualification and quality assurance must be performed with phantoms and that the same field strength must be used, but that the rest of the system setup and calibration methods could be determined by individual sites; (iii) the same pulse sequence and reconstruction methods were preferable, but the exact choice should be governed by the anatomical target; (iv) normalized metabolite area-under-curve (AUC) values and metabolite AUC were the preferred metabolism metrics. The work confirmed areas of consensus for multi-center study conduct and identified where further research is required to ascertain best practice.

MRI can provide localized assessment of lung function for monitoring people with lung disease. Hyperpolarized 129Xe MRI directly images pulmonary gas distribution but requires specialized hardware. Conventional 1H MRI acquisitions can also provide functional maps using free-breathing approaches. The purpose of this study is to evaluate regional ventilation derived from 3D ultrashort echo-time (UTE) 1H MRI using Motion-Compensated Low-Rank constrained reconstruction (MoCoLoR), by comparing against 129Xe MRI and pulmonary function testing as reference-standard. The study is retrospective in design. The study included 57 participants (25.4 ± 15.8 years, 35 males and 22 females): 12 healthy volunteers, 20 pediatric, and 25 adult people with cystic fibrosis (CF) scanned between January 2022 and February 2023. Field strength/sequence: 3T; 129Xe: 2D multislice spoiled gradient-recalled sequence; UTE 1H: variable-density 3D radial sequence. K-means-based 129Xe ventilation defect percent (VDP), forced expiratory volume in 1 s (FEV1), and lung clearance index (LCI) were evaluated against UTE 1H VDP from a modified k-means method. The correspondence of ventilation defect maps from 129Xe and UTE 1H was also evaluated. Statistical tests included the Pearson correlation coefficient (r) and t tests, with p < 0.05 considered significant. 129Xe and UTE 1H VDP were significantly correlated (r = 0.64, p =  9.1 × 10 - 8 $$ 9.1\times {10}^{-8} $$ ). Bland-Altman analysis showed a bias of -0.05 (p =  7.2 × 10 - 5 $$ 7.2\times {10}^{-5} $$ ) and limits of agreement of (0.07, -0.17). The Dice spatial accuracy of the UTE-based ventilation defect regions using 129Xe as reference was 0.64 ± 0.05. UTE 1H VDP was significantly correlated with FEV1 (r = -0.54, p =  2.9 × 10 - 4 $$ 2.9\times {10}^{-4} $$ ) and LCI (r = 0.48, p =  5.9 × 10 - 3 $$ 5.9\times {10}^{-3} $$ ) and was significantly different between healthy and CF participants (p = 0.017), although the correlations and differences were stronger for 129Xe VDP. UTE 1H VDP correlated with 129Xe VDP, FEV1, and LCI, and demonstrated high, consistent Dice spatial accuracy against 129Xe VDP. UTE 1H VDP captured variations in lung ventilation and has the advantage that it can be widely implemented on any MR system for evaluation and monitoring of patients with lung disease.

The purpose of this research is to estimate sensitivity maps when imaging X-nuclei that may not have a significant presence throughout the field of view. We propose to estimate the coil's sensitivities by solving a least-squares problem where each row corresponds to an individual estimate of the sensitivity for a given voxel. Multiple estimates come from the multiple bins of the spectrum with spectroscopy, multiple times with dynamic imaging, or multiple frequencies when utilizing spectral excitation. The method presented in this manuscript, called the L2 optimal method, is compared to the commonly used RefPeak method which uses the spectral bin with the highest energy to estimate the sensitivity maps. The L2 optimal method yields more accurate sensitivity maps when imaging a numerical phantom and is shown to yield a higher signal-to-noise ratio when imaging the brain, pancreas, and heart with hyperpolarized pyruvate as the contrast agent with hyperpolarized MRI. The L2 optimal method is able to better estimate the sensitivity by extracting more information from the measurements.

RATIONALE The etiology of pectus malformation is poorly understood, and treatment decisions are often based on incomplete evaluation. This study aims to develop and apply 3D UTE and 2D dynamic MRI-based anatomical and functional lung imaging techniques to improve our understanding of pectus malformation and to provide better measures for making treatment decisions. METHODS 3 pectus excavatum (PE) patients, 1 pectus carinatum (PC) patient, and 1 healthy volunteer were included. All images were acquired on a 3T clinical MR scanner with a 3D radial UTE sequence and a fast 2D spoiled gradient echo sequence. Ventilations and chest wall motions are compared (i) between PE, PC, and healthy subjects and (ii) between PE patients before and after being treated by a suction device1 that elevates the chest.The acquired 3D UTE raw data was reconstructed into 6 motion states by Motion-Compensated Low-Rank reconstruction (MoCoLoR)2. With the end-of-expiration being the reference state, a 3D quantitative regional ventilation map of each motion state was generated based on the Jacobian determinants of the MoCoLoR-reconstructed images3. For the 2D dynamic dataset, three points on the chest wall were temporally tracked to evaluate chest wall motion. RESULTS The maps of the PE patients show relatively large heterogeneities in the ventilation maps, with some regions showing severe lack of ventilation compared to the PC patient and a healthy volunteer (Figure 1a), who both have uniform ventilation across the lungs. Figure 1b exhibits elevated ventilation levels at the apex and anteriorly after the suction device was attached to the patient. Figures 1c and 1d show that the PE patient without the suction device experiences abnormal paradoxical chest wall motion where the two sides of the chest are elevated to a much larger extent compared to the sternum, most visible during deep breaths. By contrast, the chest motion of the PE patient with the suction device attached became more consistent between sternum and chest, resembling the motion of the healthy volunteer. CONCLUSION UTE MRI-based lung ventilation maps provide a quantitative, radiation-free way to evaluate volumetric lung ventilation and dynamic 2D MRI can be used to monitor chest wall motion. MRI is more desirable for lung imaging than CT in pediatric pectus malformation patients due to the ionizing radiation reduction and can provide novel lung function information. MRI results clearly demonstrate elevation of the anterior PE chest using a suction device improves the ventilation of the lung parenchyma.

Background: Most of the existing hyperpolarized (HP) 13C MRI analyses use univariate rate maps of pyruvate-to-lactate conversion (kPL), and radiomic-style multiparametric models extracting complex, higher-order features remain unexplored. Purpose: To establish a multivariate framework based on whole abdomen/pelvis HP 13C-pyruvate MRI and evaluate the association between multiparametric features of metabolism (MFM) and clinical outcome measures in advanced and metastatic prostate cancer. Methods: Retrospective statistical analysis was performed on 16 participants with metastatic or local-regionally advanced prostate cancer prospectively enrolled in a tertiary center who underwent HP-pyruvate MRI of abdomen or pelvis between November 2020 and May 2023. Five patients were hormone-sensitive and eleven were castration-resistant. GMP-grade [1-13C]pyruvate was polarized using a 5T clinical-research DNP polarizer, and HP MRI used a set of flexible vest-transmit, array-receive coils, and echo-planar imaging sequences. Three basic metabolic maps (kPL, pyruvate summed-over-time, and mean pyruvate time) were created by semi-automatic segmentation, from which 316 MFMs were extracted using an open-source, radiomic-compliant software package. Univariate risk classifier was constructed using a biologically meaningful feature (kPL,median), and the multivariate classifier used a two-step feature selection process (ranking and clustering). Both were correlated with progression-free survival (PFS) and overall survival (OS) (median follow-up = 22.0 months) using Cox proportional hazards model. Results: In the univariate analysis, patients harboring tumors with lower-kPL,median had longer PFS (11.2 vs. 0.5 months, p < 0.01) and OS (NR vs. 18.4 months, p < 0.05) than their higher-kPL,median counterparts. Using a hypothesis-generating, age-adjusted multivariate risk classifier, the lower-risk subgroup also had longer PFS (NR vs. 2.4 months, p < 0.002) and OS (NR vs. 18.4 months, p < 0.05). By contrast, established laboratory markers, including PSA, lactate dehydrogenase, and alkaline phosphatase, were not significantly associated with PFS or OS (p > 0.05). Key limitations of this study include small sample size, retrospective study design, and referral bias. Conclusions: Risk classifiers derived from select multiparametric HP features were significantly associated with clinically meaningful outcome measures in this small, heterogeneous patient cohort, strongly supporting further investigation into their prognostic values.

PURPOSE:Create vendor-neutral modular sequences for X-nuclear acquisitions and build an X-nuclear-enabled Pulseq interpreter for GE (GE HealthCare, Waukesha, WI) scanners. METHODS:We designed a modular 2D gradient echo spiral sequence to support several sequence formats and a modular metabolite-specific 3D balanced steady-state free precession sequence for hyperpolarized (HP) carbon-13 (13C) MRI. In addition, we developed a new Pulseq interpreter for GE scanners, named TOPPE MNS (TOPPE Multi-Nuclear Spectroscopy), to implement X-nuclear acquisitions capabilities. We evaluated TOPPE MNS and the modular sequences through phantom studies using phosphorus-31 (31P), hydrogen-2 (2H), and 13C coils, and in vivo studies including a human brain deuterium metabolic imaging study at natural abundance, HP 13C animal studies, and human renal studies. RESULTS:Data from the 13C phantom showed the accuracy of designed modular sequences and consistent performance with the product sequences. 31P, 2H, and 13C phantom studies and a multi-vendor/multi-version 13C phantom study showed accurate excitation and spatial encoding functionalities. A 2H-MRS brain volunteer study, HP [1-13C]pyruvate animal study, and human renal study showed good image quality with SNR comparable to those reported in the published literature. These results demonstrated the reproducibility of the TOPPE MNS GE interpreter and modular spiral sequences. CONCLUSION:We have designed a modular 2D gradient echo spiral sequence supporting several sequence formats and a modular metabolic-specific 3D balanced steady-state free precession sequence for 13C acquisition, as well as developed a GE interpreter with X-nucleus capabilities. Our work paves the way for future multi-site studies with acquisitions for X-nuclei across MRI vendors and software versions.

PurposeAccurate quantification of metabolism in hyperpolarized (HP) 13C MRI is essential for clinical applications. However, kinetic model parameters are often confounded by uncertainties in radiofrequency flip angles and other model parameters.MethodsA data-driven kinetic fitting approach for HP 13C-pyruvate MRI was proposed that compensates for uncertainties in the B1+ field. We hypothesized that introducing a scaling factor to the flip angle to minimize fit residuals would allow more accurate determination of the pyruvate-to-lactate conversion rate (kPL). Numerical simulations were performed under different conditions (flip angle, kPL, and T1 relaxation), with further testing using HP 13C-pyruvate MRI of rat liver and kidneys.ResultsSimulations showed that the proposed method reduced kPL error from 60% to 1% when the prescribed and actual flip angles differed by 60%. The method also showed robustness to T1 uncertainties, achieving median kPL errors within +/- 3% even when the assumed T1 was incorrect by up to a factor of 2. In rat studies, better-quality fitting for lactate signals (a 1.4-fold decrease in root mean square error [RMSE] for lactate fit) and tighter kPL distributions (an average of 3.1-fold decrease in kPL standard deviation) were achieved using the proposed method compared with when no correction was applied.ConclusionThe proposed data-driven kinetic fitting approach provided a method to accurately quantify HP 13C-pyruvate metabolism in the presence of B1+ inhomogeneity. This model may also be used to correct for other error sources, such as T1 relaxation and flow, and may prove to be clinically valuable in improving tumor staging or assessing treatment response.

Fitting rate constants to Hyperpolarized [1-13C]Pyruvate (HP C13) MRI data is a promising approach for quantifying metabolism in vivo. Current methods typically fit each voxel of the dataset using a least-squares objective. With these methods, each voxel is considered independently, and the spatial relationships are not considered during fitting.In this work, we use a convolutional neural network, a U-Net, with convolutions across the 2D spatial dimensions to estimate pyruvate-to-lactate conversion rate, kPL, maps from dynamic HP C13 datasets. We designed a framework for creating simulated anatomically accurate brain data that matches typical HP C13 characteristics to provide large amounts of data for training with ground truth results. The U-Net is initially trained with the digital phantom data and then further trained with in vivo datasets for regularization.In simulation where ground-truth kPL maps are available, the U-Net outperforms voxel-wise fitting with and without spatiotemporal denoising, particularly for low SNR data. In vivo data was evaluated qualitatively, as no ground truth is available, and before regularization the U-Net predicted kPL maps appear oversmoothed. After further training with in vivo data, the resulting kPL maps appear more realistic.This study demonstrates how to use a U-Net to estimate rate constant maps for HP C13 data, including a comprehensive framework for generating a large amount of anatomically realistic simulated data and an approach for regularization. This simulation and architecture provide a foundation that can be built upon in the future for improved performance.

Non-invasive molecular imaging methods capable of assessing tumor biology in-vivo were investigated to improve the clinical management of patients with glioma. In this study, we developed two hyperpolarized (HP) 13C-MRI techniques for clinical translation: one to evaluate blood-brain barrier (BBB) integrity and another to assess isocitrate dehydrogenase (IDH) mutation status in glioma. To evaluate BBB disruption, HP [13C,15N2]urea probe was developed, exploiting urea’s small molecular weight—approximately 15 times smaller than gadolinium-contrast agents—and its inability to cross an intact BBB. A dynamic 3D balanced SSFP acquisition was performed in healthy volunteers following intravenous injection of HP urea solution, enabling high-SNR visualization of arterial, capillary, and venous compartments. Quantitative analysis of vascular transit and spatial distribution established normative references for evaluating BBB integrity. HP 13C-urea MRI may offer superior sensitivity compared to current gadolinium-based methods by directly detecting subtle BBB disruptions as a positive-contrast signal within brain parenchyma, without requiring gadolinium administration, particularly in non-enhancing brain tumors. To evaluate the IDH mutation status of glioma, we developed HP [1-13C]alpha-ketoglutarate (aKG) MRI to monitor the metabolic reprogramming specific to this type of tumor, which involves the conversion of aKG to the oncometabolite 2-hydroxyglutarate (2HG). A dynamic 13C MRS utilizing a spectral-spatial RF pulse to independently excite aKG and its downstream metabolites was acquired following intravenous injection of HP aKG solution. Initial studies with HP [1-13C]aKG in healthy volunteers demonstrated safety and feasibility, showing glutamate production consistent with normal IDH activity. Subsequent studies in patients with IDH-mutant glioma revealed signals consistent with 2HG, suggesting feasibility for directly assessing mutant IDH activity in-vivo. Further validation is underway. Together, these developments highlight the significant clinical-research potential of HP 13C MRI for probing tumor vasculature and IDH-driven metabolism, offering complementary non-invasive biomarkers for improved diagnosis, treatment monitoring, and therapeutic stratification in glioma.

This study developed a new approach to produce sterile, hyperpolarized [13C,15N2]urea as a novel molecular imaging probe and applied it for first-ever healthy brain volunteer studies. Hyperpolarized [13C,15N2]urea, as a small, metabolically inert molecule, offers significant advantages for perfusion imaging due to its endogenous nature and excellent safety profile. The developed methods achieved a hyperpolarized [13C,15N2]urea solution (132 ± 6 mM) with 27.4 ± 5.6% polarization and a T1 = 50.4 ± 0.2 s. In healthy brain volunteer studies, high-resolution 13C imaging captured blood flow with a spatial resolution of 7.76 × 7.76 × 15 (or 10) mm3 over ~1 min following hyperpolarized [13C,15N2]urea injection, visualizing detailed vascular structures. Time-to-peak and centroid analyses showed consistent arterial and venous signal patterns across subjects. Findings suggest hyperpolarized [13C,15N2]urea may have applications beyond brain imaging, including the non-invasive perfusion assessment in various organs, cancer microenvironment, and renal function, paving the way for clinical translation.

Purpose: This study leverages the echo planar time-resolved imaging (EPTI) concept in MR fingerprinting (MRF) framework for a new time-resolved MRF (TRMRF) approach, and explores its capability for fast simultaneous quantification of multiple MR parameters including T-1, T-2, T-2*, proton density, off resonance, and B-1(+). Methods: The proposed TRMRF method uses the concept of EPTI to track the signal change along the EPI echo train for T-2* weighting with a k-t Poisson-based sampling order designed for acquisition. A two-dimensional decomposition algorithm was designed for the image reconstruction, enabling fast and precise subspace modeling. The accuracy of proposed method was evaluated by a T-1/T-2 phantom. The feasibility was demonstrated through 5 healthy volunteer brain studies. Results: In the phantom studies, T-1, T-2, and T-2* maps of TRMRF correlated strongly with gold-standard methods. The concordance correlation coefficients are 0.9999, 0.9984 and 0.9978, and R(2)s are 0.9998, 0.9971, and 0.9983. In the in vivo studies, quantitative maps were acquired with 5 healthy volunteers. TRMRF was demonstrated to have comparable results with spiral MRF and gradient-echo EPTI. TRMRF scans using 16, 10, and 6s per slice were also evaluated to demonstrate the capability of shorter scan times. Conclusion: A new approach is proposed to exploit the advantage of EPTI in the MRF framework. We demonstrate in phantom and in vivo experiments that T-1, T-2, T-2*, proton density, off resonance, and B-1(+) can be simultaneously quantified within 6s/slice by TRMRF.

Hyperpolarized carbon-13 (HP-13C) MRI enables the real-time measurement of dynamic metabolism by utilizing molecular probes whose magnetization has been transiently enhanced via dynamic nuclear polarization of 13C labels. Based on preclinical and clinical investigations demonstrating Warburg-related metabolic dysfunction and tricarboxylic acid (TCA)-cycle alterations in gliomas, HP-13C techniques appear very promising for overcoming conventional challenges to evaluating tumor burden and extent, early therapeutic response, and progression among patients noninvasively. This article surveys the multifaceted translational development of HP-13C MRI in the context of glioma imaging, while emphasizing innovation concerning the pharmacy production of hyperpolarized (HP) probes-[1-13C]/[2-13C]-pyruvate and [1-13C,5-12C]-α-ketoglutarate-that serve as nonradioactive metabolic contrast agents. Borrowing from practical experience, we present specific probe indications for isocitrate dehydrogenase (IDH)-wild-type glioblastomas and IDH-mutant gliomas together with example data to show the targeted, pathway-dependent function of these agents and their utility. Additional information pertaining to HP-13C hardware, acquisition, and postprocessing techniques provides an overview of the imaging methodology as it is currently performed at a leading institution. Considering the developing markers for progressive disease in glioblastomas and rapidly advancing capability, this unique imaging technology appears poised for translational impact following further evaluation.

In the original publication [...].

In this intercontinental, confirmatory study, we include a retrospective cohort of 22,481 MRI examinations (21,288 patients; 46 cities in 22 countries) to train and externally validate the PI-CAI-2B model, i.e., an efficient, next-generation iteration of the state-of-the-art AI system that was developed for detecting Gleason grade group ≥2 prostate cancer on MRI during the PI-CAI study. Of these examinations, 20,471 cases (19,278 patients; 26 cities in 14 countries) from two EU Horizon projects (ProCAncer-I, COMFORT) and 12 independent centers based in Europe, North America, Asia and Africa, are used for training and internal testing. Additionally, 2010 cases (2010 patients; 20 external cities in 12 countries) from population-based screening (STHLM3-MRI, IP1-PROSTAGRAM trials) and primary diagnostic settings (PRIME trial) based in Europe, North and South Americas, Asia and Australia, are used for external testing. Primary endpoint is the proportion of AI-based assessments in agreement with the standard of care diagnoses (i.e., clinical assessments made by expert uropathologists on histopathology, if available, or at least two expert urogenital radiologists in consensus; with access to patient history and peer consultation) in the detection of Gleason grade group ≥2 prostate cancer within the external testing cohorts. Our statistical analysis plan is prespecified with a hypothesis of diagnostic interchangeability to the standard of care at the PI-RADS ≥3 (primary diagnosis) or ≥4 (screening) cut-off, considering an absolute margin of 0.05 and reader estimates derived from the PI-CAI observer study (62 radiologists reading 400 cases). Secondary measures comprise the area under the receiver operating characteristic curve (AUROC) of the AI system stratified by imaging quality, patient age and patient ethnicity to identify underlying biases (if any).

Non-invasive prostate cancer classification from MRI has the potential to revolutionize patient care by providing early detection of clinically significant disease, but has thus far shown limited positive predictive value. To address this, we present a image-based deep learning method to predict clinically significant prostate cancer from screening MRI in patients that subsequently underwent biopsy with results ranging from benign pathology to the highest grade tumors. Specifically, we demonstrate that mixed supervision via diverse histopathological ground truth improves classification performance despite the cost of reduced concordance with image-based segmentation. Where prior approaches have utilized pathology results as ground truth derived from targeted biopsies and whole-mount prostatectomy to strongly supervise the localization of clinically significant cancer, our approach also utilizes weak supervision signals extracted from nontargeted systematic biopsies with regional localization to improve overall performance. Our key innovation is performing regression by distribution rather than simply by value, enabling use of additional pathology findings traditionally ignored by deep learning strategies. We evaluated our model on a dataset of 973 (testing n = 198) multi-parametric prostate MRI exams collected at UCSF from 2016-2019 followed by MRI/ultrasound fusion (targeted) biopsy and systematic (nontargeted) biopsy of the prostate gland, demonstrating that deep networks trained with mixed supervision of histopathology can feasibly exceed the performance of the Prostate Imaging-Reporting and Data System (PI-RADS) clinical standard for prostate MRI interpretation (71.6% vs 66.7% balanced accuracy and 0.724 vs 0.716 AUC).

PURPOSE:Recent work has shown MRI is able to measure and quantify signals of phospholipid membrane-bound protons associated with myelin in the human brain. This work seeks to develop an improved technique for characterizing this brain ultrashort- T 2 ∗ $$ {\mathrm{T}}_2\ast $$ component in vivo accounting for T 1 $$ {\mathrm{T}}_1 $$ weighting. METHODS:Data from ultrashort echo time scans from 16 healthy volunteers with variable flip angles (VFA) were collected and fitted into an advanced regression model to quantify signal fraction, relaxation time, and frequency shift of the ultrashort- T 2 ∗ $$ {\mathrm{T}}_2\ast $$ component. RESULTS:The fitted components show intra-subject differences of different white matter structures and significantly elevated ultrashort- T 2 ∗ $$ {\mathrm{T}}_2\ast $$ signal fraction in the corticospinal tracts measured at 0.09 versus 0.06 in other white matter structures and significantly elevated ultrashort- T 2 ∗ $$ {\mathrm{T}}_2\ast $$ frequency shift in the body of the corpus callosum at - $$ - $$ 1.5 versus - $$ - $$ 2.0 ppm in other white matter structures. CONCLUSION:The significantly different measured components and measured T 1 $$ {\mathrm{T}}_1 $$ relaxation time of the ultrashort- T 2 ∗ $$ {\mathrm{T}}_2\ast $$ component suggest that this method is picking up novel signals from phospholipid membrane-bound protons.