Low-field magnetic resonance imaging of roots in intact clayey and silty soils.

The development of a robust method to non-invasively visualize root morphology in natural soils has been hampered by the opaque, physical, and structural properties of soils. In this work we describe a novel technology, low field magnetic resonance imaging (LF-MRI), for imaging energy sorghum (Sorghum bicolor (L.) Moench) root morphology and architecture in intact soils. The use of magnetic fields much weaker than those used with traditional MRI experiments reduces the distortion due to magnetic material naturally present in agricultural soils. A laboratory based LF-MRI operating at 47 mT magnetic field strength was evaluated using two sets of soil cores: 1) soil/root cores of Weswood silt loam (Udifluventic Haplustept) and a Belk clay (Entic Hapluderts) from a conventionally tilled field, and 2) soil/root cores from rhizotrons filled with either a Houston Black (Udic Haplusterts) clay or a sandy loam purchased from a turf company. The maximum soil water nuclear magnetic resonance (NMR) relaxation time T2 (4 ms) and the typical root water relaxation time T2 (100 ms) are far enough apart to provide a unique contrast mechanism such that the soil water signal has decayed to the point of no longer being detectable during the data collection time period. 2-D MRI projection images were produced of roots with a diameter range of 1.5-2.0 mm using an image acquisition time of 15 min with a pixel resolution of 1.74 mm in four soil types. Additionally, we demonstrate the use of a data-driven machine learning reconstruction approach, Automated Transform by Manifold Approximation (AUTOMAP) to reconstruct raw data and improve the quality of the final images. The application of AUTOMAP showed a SNR (Signal to Noise Ratio) improvement of two fold on average. The use of low field MRI presented here demonstrates the possibility of applying low field MRI through intact soils to root phenotyping and agronomy to aid in understanding of root morphology and the spatial arrangement of roots in situ.

Enzymatically Shifting Nitroxides for EPR Spectroscopy and Overhauser-Enhanced Magnetic Resonance Imaging.

In vivo investigations of enzymatic processes using non-invasive approaches are a long-lasting challenge. Recently, we showed that Overhauser-enhanced MRI is suitable to such a purpose. A ß-phosphorylated nitroxide substrate prototype exhibiting keto-enol equilibrium upon enzymatic activity has been prepared. Upon enzymatic hydrolysis, a large variation of the phosphorus hyperfine coupling constant (Δa(P)=4 G) was observed. The enzymatic activities of several enzymes were conveniently monitored by electronic paramagnetic resonance (EPR). Using a 0.2 T MRI machine, in vitro and in vivo OMRI experiments were successfully performed, affording a 1200% enhanced MRI signal in vitro, and a 600% enhanced signal in vivo. These results highlight the enhanced imaging potential of these nitroxides upon specific enzymatic substrate-to-product conversion.

Alkoxyamines: toward a new family of theranostic agents against cancer.

Theranostics combines therapeutic and diagnostic or drug deposition monitoring abilities of suitable molecules. Here we describe the first steps of building an alkoxyamine-based theranostic agent against cancer. The labile alkoxyamine ALK-1 (t(1/2) = 50 min at 37 °C) cleaves spontaneously to generate (1) a highly reactive free alkyl radical used as therapeutic agents to induce cell damages leading to cell death and (2) a stable nitroxide used as contrast agent for Overhauser-enhanced magnetic resonance imaging (OMRI). The ALK-1 toxicity was studied extensively in vitro on the glioblastoma cell line U87-MG. Cell viability appeared to be dependent on ALK-1 concentration and on the time of the observation following alkoxyamine treatment. For instance, the LC50 at 72 h was 250 µM. Data showed that cell toxicity was specifically due to the in situ released alkyl radical. This radical induced oxidative stress, mitochondrial changes, and ultimately the U87 cell apoptosis. The nitroxide production, during the alkoxyamine homolysis, was monitored by OMRI, showing a progressive MRI signal enhancement to 6-fold concomitant to the ALK-1 homolysis. In conclusion, we have demonstrated for the first time that the alkoxyamines are promising molecules to build theranostic tools against solid tumors.

Uncertainty Estimation and Out-of-Distribution Detection for Deep Learning-Based Image Reconstruction using the Local Lipschitz.

Accurate image reconstruction is at the heart of diagnostics in medical imaging. Supervised deep learning-based approaches have been investigated for solving inverse problems including image reconstruction. However, these trained models encounter unseen data distributions that are widely shifted from training data during deployment. Therefore, it is essential to assess whether a given input falls within the training data distribution. Current uncertainty estimation approaches focus on providing an uncertainty map to radiologists, rather than assessing the training distribution fit. In this work, we propose a method based on the local Lipschitz metric to distinguish out-of-distribution images from in-distribution with an area under the curve of 99.94% for True Positive Rate versus False Positive Rate. We demonstrate a very strong relationship between the local Lipschitz value and mean absolute error (MAE), supported by a Spearman's rank correlation coefficient of 0.8475, to determine an uncertainty estimation threshold for optimal performance. Through the identification of false positives, we demonstrate the local Lipschitz and MAE relationship can guide data augmentation and reduce uncertainty. Our study was validated using the AUTOMAP architecture for sensor-to-image Magnetic Resonance Imaging (MRI) reconstruction. We demonstrate our approach outperforms baseline techniques of Monte-Carlo dropout and deep ensembles as well as the state-of-the-art Mean Variance Estimation (MVE) network approach. We expand our application scope to MRI denoising and Computed Tomography (CT) sparse-to-full view reconstructions using UNET architectures. We show our approach is applicable to various architectures and applications, especially in medical imaging, where preserving diagnostic accuracy of reconstructed images remains paramount.

Breast imaging with an ultra-low field MRI scanner: a pilot study.

Breast cancer screening is necessary to reduce mortality due to undetected breast cancer. Current methods have limitations, and as a result many women forego regular screening. Magnetic resonance imaging (MRI) can overcome most of these limitations, but access to conventional MRI is not widely available for routine annual screening. Here, we used an MRI scanner operating at ultra-low field (ULF) to image the left breasts of 11 women (mean age, 35 years ±13 years) in the prone position. Three breast radiologists reviewed the imaging and were able to discern the breast outline and distinguish fibroglandular tissue (FGT) from intramammary adipose tissue. Additionally, the expert readers agreed on their assessment of the breast tissue pattern including fatty, scattered FGT, heterogeneous FGT, and extreme FGT. This preliminary work demonstrates that ULF breast MRI is feasible and may be a potential option for comfortable, widely deployable, and low-cost breast cancer diagnosis and screening.

Real-time radial reconstruction with domain transform manifold learning for MRI-guided radiotherapy.

BACKGROUND: MRI-guidance techniques that dynamically adapt radiation beams to follow tumor motion in real time will lead to more accurate cancer treatments and reduced collateral healthy tissue damage. The gold-standard for reconstruction of undersampled MR data is compressed sensing (CS) which is computationally slow and limits the rate that images can be available for real-time adaptation. PURPOSE: Once trained, neural networks can be used to accurately reconstruct raw MRI data with minimal latency. Here, we test the suitability of deep-learning-based image reconstruction for real-time tracking applications on MRI-Linacs. METHODS: We use automated transform by manifold approximation (AUTOMAP), a generalized framework that maps raw MR signal to the target image domain, to rapidly reconstruct images from undersampled radial k-space data. The AUTOMAP neural network was trained to reconstruct images from a golden-angle radial acquisition, a benchmark for motion-sensitive imaging, on lung cancer patient data and generic images from ImageNet. Model training was subsequently augmented with motion-encoded k-space data derived from videos in the YouTube-8M dataset to encourage motion robust reconstruction. RESULTS: AUTOMAP models fine-tuned on retrospectively acquired lung cancer patient data reconstructed radial k-space with equivalent accuracy to CS but with much shorter processing times. Validation of motion-trained models with a virtual dynamic lung tumor phantom showed that the generalized motion properties learned from YouTube lead to improved target tracking accuracy. CONCLUSION: AUTOMAP can achieve real-time, accurate reconstruction of radial data. These findings imply that neural-network-based reconstruction is potentially superior to alternative approaches for real-time image guidance applications.

ARmedViewer, an augmented-reality-based fast 3D reslicer for medical image data on mobile devices: A feasibility study.

BACKGROUND AND OBJECTIVE: Medical images obtained by methods such as magnetic resonance imaging (MRI) or computed tomography (CT) are typically displayed as a stack of 2D slices, comprising a 3D volume. Often, the anatomy of interest does not fall neatly into the slice plane but rather extends obliquely through several slices. Reformatting the data to show the anatomy in one slice in conventional medical imaging software can require expertise and time. In this work, we present ARmedViewer, a medical image viewing app designed for mobile devices that uses augmented reality technology to display medical image data. An arbitrary plane for displaying the data can be chosen quickly and intuitively by moving the mobile device. METHODS: The app ARmedViewer, compiled for an iOS device, was designed to allow a user to easily select from a list of 3D image datasets consisting of header information and image data. The user decides where to place the data, which can be overlaid on actual human anatomy. After loading the dataset, the user can move and rotate the data as desired. 15 users compared the user experience of the app to a common image viewer by answering two user surveys each, one custom and one standardized. The utility of the app was also tested by having two users find a plane through a 3D dataset that displayed 3 randomly placed lesions. This operation was timed and compared between the app and a standard medical image viewer. RESULTS: ARmedViewer was successfully developed and run on an iPhone XS. User interfaces for selecting, placing, moving, reslicing, and displaying the data were operated with ease, even by naïve users. The custom user survey indicated that freely selecting a slice through the data was significantly more intuitive and easier using the app than using a conventional image viewer on a computer workstation, and changing the viewing angle was also significantly more intuitive. The standardized survey indicated a significantly better user experience for the app in several categories, and never significantly worse. The timed reslicing experiments demonstrated the app being faster than the standard image viewer by an average factor of 9. CONCLUSIONS: The newly developed ARmedViewer is a portable software tool for easily displaying 3D medical image data overlaid on human anatomy, allowing for easy choice of the viewing plane by intuitively moving the mobile device.

Optimization of a Close-Fitting Volume RF Coil for Brain Imaging at 6.5 mT Using Linear Programming.

OBJECTIVE: The use of a close-fitting roughly head-shaped volume coil for MRI (magnetic resonance imaging) has the merit of improving the filling factor and thus the SNR (signal-to-noise ratio) from the brain. However, the surface of the RF coil follows that of the head which makes it difficult to determine an optimal coil winding pattern. We describe here a new method to optimize a head-shaped RF coil with the objective of maximizing its SNR and RF-magnetic-field homogeneity for operation at ultra-low magnetic field (6.5 mT, 276 kHz). METHODS: The approach consists of FEM (finite-element-method) simulation and linear programing based optimization. RESULTS: We have implemented the optimization and further studied the relationship between the design requirements and the performance of the RF coil. Finally, we constructed an optimal RF coil and scanned both a head-shaped phantom and a human subject. CONCLUSION: The method we outline here provide new insight into the conductor layout needed for magnetic optimization of structurally complex coils, especially when tradeoffs between competing attributes (SNR and homogeneity in this case) must be made.

Synthesizing Quantitative T2 Maps in Right Lateral Knee Femoral Condyles from Multicontrast Anatomic Data with a Conditional Generative Adversarial Network.

PURPOSE: To develop a proof-of-concept convolutional neural network (CNN) to synthesize T2 maps in right lateral femoral condyle articular cartilage from anatomic MR images by using a conditional generative adversarial network (cGAN). MATERIALS AND METHODS: In this retrospective study, anatomic images (from turbo spin-echo and double-echo in steady-state scans) of the right knee of 4621 patients included in the 2004-2006 Osteoarthritis Initiative were used as input to a cGAN-based CNN, and a predicted CNN T2 was generated as output. These patients included men and women of all ethnicities, aged 45-79 years, with or at high risk for knee osteoarthritis incidence or progression who were recruited at four separate centers in the United States. These data were split into 3703 (80%) for training, 462 (10%) for validation, and 456 (10%) for testing. Linear regression analysis was performed between the multiecho spin-echo (MESE) and CNN T2 in the test dataset. A more detailed analysis was performed in 30 randomly selected patients by means of evaluation by two musculoskeletal radiologists and quantification of cartilage subregions. Radiologist assessments were compared by using two-sided t tests. RESULTS: The readers were moderately accurate in distinguishing CNN T2 from MESE T2, with one reader having random-chance categorization. CNN T2 values were correlated to the MESE values in the subregions of 30 patients and in the bulk analysis of all patients, with best-fit line slopes between 0.55 and 0.83. CONCLUSION: With use of a neural network-based cGAN approach, it is feasible to synthesize T2 maps in femoral cartilage from anatomic MRI sequences, giving good agreement with MESE scans.See also commentary by Yi and Fritz in this issue.Keywords: Cartilage Imaging, Knee, Experimental Investigations, Quantification, Vision, Application Domain, Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms© RSNA, 2021.

Detection of nanotesla AC magnetic fields using steady-state SIRS and ultra-low field MRI.

OBJECTIVE: Functional magnetic resonance imaging (fMRI) is commonly used to measure brain activity through the blood oxygen level dependent (BOLD) signal mechanism, but this only provides an indirect proxy signal to neuronal activity. Magnetoencephalography (MEG) provides a more direct measurement of the magnetic fields created by neuronal currents in the brain, but requires very specialized hardware and only measures these fields at the scalp. Recently, progress has been made to directly detect neuronal fields with MRI using the stimulus-induced rotary saturation (SIRS) effect, but interference from the BOLD response complicates such measurements. Here, we describe an approach to detect nanotesla-level, low-frequency alternating magnetic fields with an ultra-low field (ULF) MRI scanner, unaffected by the BOLD signal. APPROACH: A steady-state implementation of the stimulus-induced rotary saturation (SIRS) method is developed. The method is designed to generate a strong signal at ultra-low magnetic field as well as allowing for efficient signal averaging, giving a high contrast-to-noise ratio (CNR). The method is tested in computer simulations and in phantom scans. MAIN RESULTS: The simulations and phantom scans demonstrated the ability of the method to measure magnetic fields at different frequencies at ULF with a stronger contrast than non-steady-state approaches. Furthermore, the rapid imaging functionality of the method reduced noise efficiently. The results demonstrated sufficient CNR down to 7 nT, but the sensitivity will depend on the imaging parameters. SIGNIFICANCE: A steady-state SIRS method is able to detect low-frequency alternating magnetic fields at ultra-low main magnetic field strengths with a large signal response and contrast-to-noise, presenting an important step in sensing biological fields with ULF MRI.

3D anatomical and perfusion MRI for longitudinal evaluation of biomaterials for bone regeneration of femoral bone defect in rats.

Magnetic Resonance Imaging (MRI) appears as a good surrogate to Computed Tomography (CT) scan as it does not involve radiation. In this context, a 3D anatomical and perfusion MR imaging protocol was developed to follow the evolution of bone regeneration and the neo-vascularization in femoral bone defects in rats. For this, three different biomaterials based on Pullulan-Dextran and containing either Fucoidan or HydroxyApatite or both were implanted. In vivo MRI, ex vivo micro-CT and histology were performed 1, 3 and 5 weeks after implantation. The high spatially resolved (156 × 182 × 195 µm) anatomical images showed a high contrast from the defects filled with biomaterials that decreased over time due to bone formation. The 3D Dynamic Contrast Enhanced (DCE) imaging with high temporal resolution (1 image/19 s) enabled to detect a modification in the Area-Under-The-Gadolinium-Curve over the weeks post implantation. The high sensitivity of MRI enabled to distinguish which biomaterial was the least efficient for bone regeneration, which was confirmed by micro-CT images and by a lower vessel density observed by histology. In conclusion, the methodology developed here highlights the efficiency of longitudinal MRI for tissue engineering as a routine small animal exam.

In vivo MEMRI characterization of brain metastases using a 3D Look-Locker T1-mapping sequence.

Although MEMRI (Manganese Enhanced MRI) informations were obtained on primary tumors in small animals, MEMRI data on metastases are lacking. Thus, our goal was to determine if 3D Look-Locker T1 mapping was an efficient method to evaluate Mn ions transport in brain metastases in vivo. The high spatial resolution in 3D (156 × 156 × 218 µm) of the sequence enabled to detect metastases of 0.3 mm3. In parallel, the T1 quantitation enabled to distinguish three populations of MDA-MB-231 derived brain metastases after MnCl2 intravenous injection: one with a healthy blood-tumor barrier that did not internalize Mn2+ ions, and two others, which T1 shortened drastically by 54.2% or 24%. Subsequent scans of the mice, enabled by the fast acquisition (23 min), demonstrated that these T1 reached back their pre-injection values in 24 h. Contrarily to metastases, the T1 of U87-MG glioma remained 26.2% shorter for one week. In vitro results supported the involvement of the Transient Receptor Potential channels and the Calcium-Sensing Receptor in the uptake and efflux of Mn2+ ions, respectively. This study highlights the ability of the 3D Look-Locker T1 mapping sequence to study heterogeneities (i) amongst brain metastases and (ii) between metastases and glioma regarding Mn transport.

In vivo Overhauser-enhanced MRI of proteolytic activity.

There is an increasing interest in developing novel imaging strategies for sensing proteolytic activities in intact organisms in vivo. Overhauser-enhanced MRI (OMRI) offers the possibility to reveal the proteolysis of nitroxide-labeled macromolecules thanks to a sharp decrease of the rotational correlation time of the nitroxide moiety upon cleavage. In this paper, this concept is illustrated in vivo at 0.2 T using nitroxide-labeled elastin orally administered in mice. In vitro, this elastin derivative was OMRI-visible and gave rise to high Overhauser enhancements (19-fold at 18 mm nitroxide) upon proteolysis by pancreatic porcine elastase. In vivo three-dimensional OMRI detection of proteolysis was carried out. A keyhole fully balanced steady-state free precession sequence was used, which allowed 3D OMRI acquisition within 20 s at 0.125 mm(3) resolution. About 30 min after mouse gavage, proteolysis was detected in the duodenum, where Overhauser enhancements were 7.2 ± 2.4 (n = 7) and was not observed in the stomach. Conversely, orally administered free nitroxides or pre-digested nitroxide-labeled elastin were detected in the mouse's stomach by OMRI. Combined with specific molecular probes, this Overhauser-enhanced MRI technique can be used to evaluate unregulated proteolytic activities in various models of experimental diseases and for drug testing.

Overhauser-enhanced MRI of elastase activity from in vitro human neutrophil degranulation.

BACKGROUND: Magnetic resonance imaging can reveal exquisite anatomical details. However several diseases would benefit from an imaging technique able to specifically detect biochemical alterations. In this context protease activity imaging is one of the most promising areas of research. METHODOLOGY/PRINCIPAL FINDINGS: We designed an elastase substrate by grafting stable nitroxide free radicals on soluble elastin. This substrate generates a high Overhauser magnetic resonance imaging (OMRI) contrast upon digestion by the target proteases through the modulation of its rotational correlation time. The sensitivity is sufficient to generate contrasted images of the degranulation of neutrophils induced by a calcium ionophore from 2×10(4) cells per milliliter, well under the physiological neutrophils concentrations. CONCLUSIONS/SIGNIFICANCE: These ex-vivo experiments give evidence that OMRI is suitable for imaging elastase activity from neutrophil degranulation. Provided that a fast protease-substrate is used these results open the door to better diagnoses of a number of important pathologies (cystic fibrosis, inflammation, pancreatitis) by OMRI or Electron Paramagnetic Resonance Imaging in vivo. It also provides a long-expected method to monitor anti-protease treatments efficiency and help pharmaceutical research.