Clipped DeepControl: Deep neural network two-dimensional pulse design with an amplitude constraint layer.

Advanced radio-frequency pulse design used in magnetic resonance imaging has recently been demonstrated with deep learning of (convolutional) neural networks and reinforcement learning. For two-dimensionally selective radio-frequency pulses, the (convolutional) neural network pulse prediction time (a few milliseconds) was in comparison more than three orders of magnitude faster than the conventional optimal control computation. The network pulses were from the supervised training capable of compensating scan-subject dependent inhomogeneities of B0 and B1+ fields. Unfortunately, the network presented with a small percentage of pulse amplitude overshoots in the test subset, despite the optimal control pulses used in training were fully constrained. Here, we have extended the convolutional neural network with a custom-made clipping layer that completely eliminates the risk of pulse amplitude overshoots, while preserving the ability to compensate for the inhomogeneous field conditions.

Computer-Assisted Annotation of Digital H&E/SOX10 Dual Stains Generates High-Performing Convolutional Neural Network for Calculating Tumor Burden in H&E-Stained Cutaneous Melanoma.

Deep learning for the analysis of H&E stains requires a large annotated training set. This may form a labor-intensive task involving highly skilled pathologists. We aimed to optimize and evaluate computer-assisted annotation based on digital dual stains of the same tissue section. H&E stains of primary and metastatic melanoma (N = 77) were digitized, re-stained with SOX10, and re-scanned. Because images were aligned, annotations of SOX10 image analysis were directly transferred to H&E stains of the training set. Based on 1,221,367 annotated nuclei, a convolutional neural network for calculating tumor burden (CNNTB) was developed. For primary melanomas, precision of annotation was 100% (95%CI, 99% to 100%) for tumor cells and 99% (95%CI, 98% to 100%) for normal cells. Due to low or missing tumor-cell SOX10 positivity, precision for normal cells was markedly reduced in lymph-node and organ metastases compared with primary melanomas (p < 0.001). Compared with stereological counts within skin lesions, mean difference in tumor burden was 6% (95%CI, -1% to 13%, p = 0.10) for CNNTB and 16% (95%CI, 4% to 28%, p = 0.02) for pathologists. Conclusively, the technique produced a large annotated H&E training set with high quality within a reasonable timeframe for primary melanomas and subcutaneous metastases. For these lesion types, the training set generated a high-performing CNNTB, which was superior to the routine assessments of pathologists.

Optimal control gradient precision trade-offs: Application to fast generation of DeepControl libraries for MRI.

We have recently demonstrated supervised deep learning methods for rapid generation of radiofrequency pulses in magnetic resonance imaging (https://doi.org/10.1002/mrm.27740, https://doi.org/10.1002/mrm.28667). Unlike the previous iterative optimization approaches, deep learning methods generate a pulse using a fixed number of floating-point operations - this is important in MRI, where patient-specific pulses preferably must be produced in real time. However, deep learning requires vast training libraries, which must be generated using the traditional methods, e.g., iterative quantum optimal control methods. Those methods are usually variations of gradient descent, and the calculation of the gradient of the performance metric with respect to the pulse waveform can be the most numerically intensive step. In this communication, we explore various ways in which the calculation of gradients in quantum optimal control theory may be accelerated. Four optimization avenues are explored: truncated commutator series expansions at zeroth and first order, a novel midpoint truncation scheme at first order, and the exact complex-step method. For the spin systems relevant to MRI, the first-order midpoint truncation is found to be sufficiently accurate, but also significantly faster than the machine precision gradient. This makes the generation of training databases for the machine learning methods considerably more realistic.

DeepControl: 2DRF pulses facilitating B1+ inhomogeneity and B0 off-resonance compensation in vivo at 7 T.

PURPOSE: Rapid 2DRF pulse design with subject-specific B1+ inhomogeneity and B0 off-resonance compensation at 7 T predicted from convolutional neural networks is presented. METHODS: The convolution neural network was trained on half a million single-channel transmit 2DRF pulses optimized with an optimal control method using artificial 2D targets, B1+ and B0 maps. Predicted pulses were tested in a phantom and in vivo at 7 T with measured B1+ and B0 maps from a high-resolution gradient echo sequence. RESULTS: Pulse prediction by the trained convolutional neural network was done on the fly during the MR session in approximately 9 ms for multiple hand-drawn regions of interest and the measured B1+ and B0 maps. Compensation of B1+ inhomogeneity and B0 off-resonances has been confirmed in the phantom and in vivo experiments. The reconstructed image data agree well with the simulations using the acquired B1+ and B0 maps, and the 2DRF pulse predicted by the convolutional neural networks is as good as the conventional RF pulse obtained by optimal control. CONCLUSION: The proposed convolutional neural network-based 2DRF pulse design method predicts 2DRF pulses with an excellent excitation pattern and compensated B1+ and B0 variations at 7 T. The rapid 2DRF pulse prediction (9 ms) enables subject-specific high-quality 2DRF pulses without the need to run lengthy optimizations.

Imaging Rheumatoid Arthritis in Mice Using Combined Near Infrared and 19F Magnetic Resonance Modalities.

Rheumatoid arthritis (RA) is an autoimmune disease that causes pain and tissue destruction in people worldwide. An accurate diagnosis is paramount in order to develop an effective treatment plan. This study demonstrates that combining near infrared (NIR) imaging and 19F MRI with the injection of labelled nanoparticles provides high diagnostic specificity for RA. The nanoparticles were made from poly(ethylene glycol)-block-poly(lactic-co-glycolic acid) (NP) or PLGA-PEG-Folate (Folate-NP), loaded with perfluorooctyl bromide (PFOB) and indocyanine green (ICG) and evaluated in vitro and in a collagen-induced arthritic (CIA) mouse model. The different particles had a similar size and a spherical shape according to dynamic light scattering (DLS) and transmission electron microscopy (TEM). Based on flow cytometry and 19F MRI analysis, Folate-NP yielded a higher uptake than NP in activated macrophages in vitro. The potential RA-targeting ability of the particles was studied in CIA mice using NIR and 19F MRI analysis. Both NP and Folate-NP accumulated in the RA tissues, where they were visible in NIR and 19F MRI for up to 24 hours. The presence of folate as a targeting ligand significantly improved the NIR signal from inflamed tissue at the early time point (2 hours), but not at later time points. Overall, these results suggest that our nanoparticles can be applied for combined NIR and 19F MRI imaging for improved RA diagnosis.

Pluronic F127-Folate Coated Super Paramagenic Iron Oxide Nanoparticles as Contrast Agent for Cancer Diagnosis in Magnetic Resonance Imaging.

Contrast agents have been widely used in medicine to enhance contrast in magnetic resonance imaging (MRI). Among them, super paramagnetic iron oxide nanoparticles (SPION) have been reported to have low risk in clinical use. In our study, F127-Folate coated SPION was fabricated in order to efficiently target tumors and provide imaging contrast in MRI. SPION alone have an average core size of 15 nm. After stabilizing with Pluronic F127, the nanoparticles reached a hydrodynamic size of 180 nm and dispersed well in various kinds of media. The F127-Folate coated SPION were shown to specifically target folate receptor expressing cancer cells by flow cytometry analysis, confocal laser scanning microscope, as well as in vitro MRI. Furthermore, in vivo MRI images have shown the enhanced negative contrast from the F127-Folate coated SPION in tumor-bearing mice. In conclusion, our F127-Folate coated SPION have shown great potential as a contrast agent in MRI, as well as in the combination with drug delivery for cancer therapy.

Ultrafast (milliseconds), multidimensional RF pulse design with deep learning.

PURPOSE: Some advanced RF pulses, like multidimensional RF pulses, are often long and require substantial computation time because of a number of constraints and requirements, sometimes hampering clinical use. However, the pulses offer opportunities of reduced-FOV imaging, regional flip-angle homogenization, and localized spectroscopy, e.g., of hyperpolarized metabolites. Proposed herein is a novel deep learning approach to ultrafast design of multidimensional RF pulses with intention of real-time pulse updates. METHODS: The proposed neural network considers input maps of the desired excitation region of interest and outputs a single-channel, multidimensional RF pulse. The training library is, e.g., retrieved from a large image database, and the target RF pulses trained upon are calculated with a method of choice. RESULTS: A relatively simple neural network is enough to produce reliable 2D spatial-selective RF pulses of comparable performance to the teaching method. For binary regions of interest, the training library does not need to be vast; hence, reestablishment of the training library is not necessarily cumbersome. The predicted pulses were tested numerically and experimentally at 3 T. CONCLUSION: Relatively effortless training of multidimensional RF pulses, based on non-MRI-related inputs, but working in an MRI setting still, has been demonstrated. The prediction time of a few milliseconds renders real-time updates of advanced RF pulses possible.

Cryogen-free dissolution dynamic nuclear polarization polarizer operating at 3.35 T, 6.70 T, and 10.1 T.

PURPOSE: A novel dissolution dynamic nuclear polarization (dDNP) polarizer platform is presented. The polarizer meets a number of key requirements for in vitro, preclinical, and clinical applications. METHOD: It uses no liquid cryogens, operates in continuous mode, accommodates a wide range of sample sizes up to and including those required for human studies, and is fully automated. RESULTS: It offers a wide operational window both in terms of magnetic field, up to 10.1 T, and temperature, from room temperature down to 1.3 K. The polarizer delivers a 13 C liquid state polarization for [1-13 C]pyruvate of 70%. The build-up time constant in the solid state is approximately 1200 s (20 minutes), allowing a sample throughput of at least one sample per hour including sample loading and dissolution. CONCLUSION: We confirm the previously reported strong field dependence in the range 3.35 to 6.7 T, but see no further increase in polarization when increasing the magnetic field strength to 10.1 T for [1-13 C]pyruvate and trityl. Using a custom dry magnet, cold head and recondensing, closed-cycle cooling system, combined with a modular DNP probe, and automation and fluid handling systems, we have designed a unique dDNP system with unrivalled flexibility and performance.

Theranostic poly(lactic-co-glycolic acid) nanoparticle for magnetic resonance/infrared fluorescence bimodal imaging and efficient siRNA delivery to macrophages and its evaluation in a kidney injury model.

In this work, a theranostic nanoparticle was developed for multimodal imaging and siRNA delivery. The core of the nanoparticles (NP) was formed by encapsulation of superparamagnetic iron oxides and indocyanine green in a PLGA matrix to serve as a multimodal probe for near-infrared (NIFR) and magnetic resonance (MR) imaging. The surface of the particle was coated with polyethylenimine (PEI) for siRNA delivery. Macrophages efficiently took up the nanoparticles and emitted strong NIFR and MR contrast. When transfected with siRNA targeting the pro-inflammatory enzyme cyclooxygenase-2 (COX-2), significant down-regulation of COX-2 was achieved in activated macrophages. Furthermore, after injection into a unilateral ureteral obstruction (UUO)-induced kidney injury model, NIFR and MRI imaging revealed accumulation of nanoparticles in the injury kidney. In addition, in vivo silencing of COX-2 was achieved by NP/PEI/siCOX-2, which further attenuated kidney injury. Our theranostic platform represents a promising approach for simultaneous diagnosis and treatment of inflammatory diseases.

Theranostic tumor targeted nanoparticles combining drug delivery with dual near infrared and 19F magnetic resonance imaging modalities.

Combining imaging and drug delivery of "theranostic" nanoparticles has enabled concurrent diagnosis and therapy of diseases. Here, we describe a novel theranostic system that combines two imaging tracers, perfluorooctyl bromide (PFOB) for 19F magnetic resonance imaging (MRI) and indocyanine green (ICG) for near infrared (NIR) imaging, with the chemotherapeutic agent doxorubicin (Dox) into poly (lactic-co-glycolic acid)- poly (ethylene-glycol)-folate (PLGA-PEG-folate) nanoparticles. Cell culture studies using flow cytometry, confocal laser scanning microscope imaging, and 19F MRI showed enhanced uptake of nanoparticles via folate receptors expressed on human nasopharyngeal epidermal carcinoma (KB) cells. In vivo, higher MRI and fluorescence signals were obtained from tumors with 19F MRI and NIR, respectively, using folate-receptor-targeted nanoparticles compared with non-targeted equivalents. An in vitro cytotoxicity assay showed that folate-targeted nanoparticles were able to kill cancer cells more efficiently than non-folate conjugated particles. Our results suggest a potential use of PLGA-PEG-folate PFOB/ICG/Dox nanoparticles as a targeted chemotherapy agent traceable by either 19F MRI or NIR imaging.

Chitosan-coated poly(lactic-co-glycolic acid) perfluorooctyl bromide nanoparticles for cell labeling in (19)F magnetic resonance imaging.

Noninvasive therapeutic cell tracking methods in living animals are important for understanding cell function and fate in connection with cell therapy. Here we report a new particle system based on chitosan-coated poly(lactic-co-glycolic acid) perfluorooctyl bromide (PLGA PFOB) nanoparticles designed for (19)F magnetic resonance imaging (MRI) cell tracking. Chitosan was adsorbed onto the PLGA PFOB nanoparticles through electric interactions, which led to an increase in the hydrodynamic size and a surface charge proportional to the coating weight ratio. Confocal laser scanning microscopy, flow cytometry analysis and (19)F-MRI showed that to achieve the highest labeling efficiency in vitro, the optimal weight ratio of chitosan to the PLGA PFOB nanoparticles was 1:10 for human mesenchymal stem cells (hMSCs) and 1:100 for Raw 264.7 macrophages. In vivo(19)F-MRI showed that (19)F labeled hMSCs remained at the injected site 24h after injection. Thus, this study validates that chitosan-coated PLGA PFOB nanoparticles have the potential to track cell migration in vivo.

Storage of magnetization as singlet order by optimal control designed pulses.

PURPOSE: The use of hyperpolarization to enhance the sensitivity of MRI has so far been limited by the decay of the polarization through T1 relaxation. Recently, methods have been proposed that extend the lifetime of the hyperpolarization by storing the spin order in slowly relaxing singlet states. METHODS: With this aim, optimal control theory was applied to create pulses that for near-equivalent spins accomplish transfers in and out of the singlet state with maximum efficiency while ensuring robustness toward variations in the nuclear spin system Hamiltonian (chemical shift, J-couplings, B1 and B0 magnetic field inhomogeneity). RESULTS: The pulses are designed to accomplish efficient transfer with low B1 amplitude, essential for applications on preclinical and clinical MR scanners. CONCLUSION: It is demonstrated that significantly improved efficiency and robustness can be obtained within the limitations of typical MR scanner performance.