Purpose Calculation of sophisticated MR contrasts often requires complex mathematical modeling. Data evaluation is computationally expensive, vulnerable to artifacts, and often sensitive to fit algorithm parameters. In this work, we investigate whether neural networks can provide not only fast model fitting results, but also a quality metric for the predicted values, so called uncertainty quantification, investigated here in the context of multi‐pool Lorentzian fitting of CEST MRI spectra at 3T. Methods A deep feed‐forward neural network including a probabilistic output layer allowing for uncertainty quantification was set up to take uncorrected CEST‐spectra as input and predict 3T Lorentzian parameters of a 4‐pool model (water, semisolid MT, amide CEST, NOE CEST), including the B0 inhomogeneity. Networks were trained on data from 3 subjects with and without data augmentation, and applied to untrained data from 1 additional subject and 1 brain tumor patient. Comparison to conventional Lorentzian fitting was performed on different perturbations of input data. Results The deepCEST 3T networks provided fast and accurate predictions of all Lorentzian parameters and were robust to input perturbations because of noise or B0 artifacts. The uncertainty quantification detected fluctuations in input data by increase of the uncertainty intervals. The method generalized to unseen brain tumor patient CEST data. Conclusions The deepCEST 3T neural network provides fast and robust estimation of CEST parameters, enabling online reconstruction of sophisticated CEST contrast images without the typical computational cost. Moreover, the uncertainty quantification indicates if the predictions are trustworthy, enabling confident interpretation of contrast changes.

Purpose To determine the feasibility of employing the prior knowledge of well‐separated chemical exchange saturation transfer (CEST) signals in the 9.4 T Z‐spectrum to separate overlapping CEST signals acquired at 3 T, using a deep learning approach trained with 3 T and 9.4 T CEST spectral data from brains of the same subjects. Methods Highly spectrally resolved Z‐spectra from the same volunteer were acquired by 3D‐snapshot CEST MRI at 3 T and 9.4 T at low saturation power of B1 = 0.6 µT. The volume‐registered 3 T Z‐spectra‐stack was then used as input data for a three layer deep neural network with the volume‐registered 9.4 T fitted parameter stack as target data. Results An optimized neural net architecture could be found and verified in healthy volunteers. The gray‐/white‐matter contrast of the different CEST effects was predicted with only small deviations (Pearson R = 0.89). The 9.4 T prediction was less noisy compared to the directly measured CEST maps, although at the cost of slightly lower tissue contrast. Application to an unseen tumor patient measured at 3 T and 9.4 T revealed that tumorous tissue Z‐spectra and corresponding hyper‐/hypointensities of different CEST effects can also be predicted (Pearson R = 0.84). Conclusion The 9.4 T CEST signals acquired at low saturation power can be accurately estimated from CEST imaging at 3 T using a neural network trained with coregistered 3 T and 9.4 T data of healthy subjects. The deepCEST approach generalizes to Z‐spectra of tumor areas and might indicate whether additional ultrahigh‐field (UHF) scans will be beneficial.

Magnetic resonance (MR) images can be created noninvasively using only static and dynamic magnetic fields, and radio frequency pulses. MR imaging provides fast image acquisitions which have been clinically feasible only since the discovery of efficient MR sequences, 1-3 ie, time-efficient application of two building blocks: radio frequency pulses and spatial magnetic

Purpose To improve whole‐brain SNR at 7 Tesla, a novel 32‐element hybrid human head array coil was developed, constructed, and tested. Methods Our general design strategy is based on 2 major ideas: Firstly, following suggestions of previous works based on the ultimate intrinsic SNR theory, we combined loops and dipoles for improvement of SNR near the head center. Secondly, we minimized the total number of array elements by using a hybrid combination of transceive (TxRx) and receive (Rx) elements. The new hybrid array consisted of 8 folded‐end TxRx‐dipole antennas and 3 rows of 24 Rx‐loops all placed in a single layer on the surface of a tight‐fit helmet. Results The developed array significantly improved SNR in vivo both near the center (∼20%) and at the periphery (∼20% to 80%) in comparison to a common commercial array coil with 8 transmit (Tx) and 32 Rx‐elements. Whereas 24 loops alone delivered central SNR very similar to that of the commercial coil, the addition of complementary dipole structures provided further improvement. The new array also provided ∼15% higher Tx efficiency and better longitudinal coverage than that of the commercial array. Conclusion The developed array coil demonstrated advantages in combining complementary TxRx and Rx resonant structures, that is, TxRx‐dipoles and Rx‐loops all placed in a single layer at the same distance to the head. This strategy improved both SNR and Tx‐performance, as well as simplified the total head coil design, making it more robust.

Purpose To investigate how electronically modulated time‐varying receive sensitivities can improve parallel imaging reconstruction at ultra‐high field. Methods Receive sensitivity modulation was achieved by introducing PIN diodes in the receive loops, which allow rapid switching of capacitances in both arms of each loop coil and by that alter B1− profiles, resulting in two distinct receive sensitivity configurations. A prototype 8‐channel reconfigurable receive coil for human head imaging at 9.4T was built, and MR measurements were performed in both phantom and human subject. A modified SENSE reconstruction for time‐varying sensitivities was formulated, and g‐factor calculations were performed to investigate how modulation of receive sensitivity profiles during image encoding can improve parallel imaging reconstruction. The optimized modulation pattern was realized experimentally, and reconstructions with the time‐varying sensitivities were compared with conventional static SENSE reconstructions. Results The g‐factor calculations showed that fast modulation of receive sensitivities in the order of the ADC dwell time during k‐space acquisition can improve parallel imaging performance, as this effectively makes spatial information of both configurations simultaneously available for image encoding. This was confirmed by in vivo measurements, for which lower reconstruction errors (SSIM = 0.81 for acceleration R = 4) and g‐factors (max g = 2.4; R = 4) were observed for the case of rapidly switched sensitivities compared to conventional reconstruction with static sensitivities (SSIM = 0.74 and max g = 3.2; R = 4). As the method relies on the short RF wavelength at ultra‐high field, it does not yield significant benefits at 3T and below. Conclusions Time‐varying receive sensitivities can be achieved by inserting PIN diodes in the receive loop coils, which allow modulation of B1− patterns. This offers an additional degree of freedom for image encoding, with the potential for improved parallel imaging performance at ultra‐high field.