Apparatus for imaging during surgical procedures includes an operating room for the surgical procedure and an MRI for obtaining images periodically through the surgical procedure by moving the magnet up to the table. The magnet wire is formed of a superconducting material such as magnesium di-boride or Niobium-Titanium which is cooled by a vacuum cryocooling system to superconductivity without use of liquid helium. The magnet weighs less than 1 to 2 tonne and has a floor area in the range 15 to 35 sq feet so that it can be carried on the floor by a support system having an air cushion covering the base area of the magnet having side skirts so as to spread the weight over the entire base area. The magnet remains in the room during surgery and is powered off to turn off the magnetic field when in the second position remote from the table.

Methods and systems are provided for predicting B.sub.1.sup.+ field maps from magnetic resonance calibration images using deep neural networks. In an exemplary embodiment a method for magnetic resonance imaging comprises, acquiring a magnetic resonance (MR) calibration image of an anatomical region, mapping the MR calibration image to a transmit field map (B.sub.1.sup.+ field map) with a trained deep neural network, acquiring a diagnostic MR image of the anatomical region, and correcting inhomogeneities of a transmit field in the diagnostic MR image with the B.sub.1.sup.+ field map. Further, methods and systems are provided for collecting and processing training data, as well as utilizing the training data to train a deep learning network to predict B.sub.1.sup.+ field maps from MR calibration images.

The invention provides for a magnetic resonance imaging system (100) comprising a radio frequency system (116, 114, 118) configured for acquiring magnetic resonance data (144) from an imaging zone (108). The radio frequency system is configured for sending and receiving radio frequency signals to acquire the magnetic resonance data, wherein the radio frequency system comprises: an elliptical transmission coil (114) configured for generating a B1+ excitation field within the imaging zone; and an active B1 shim coil (118) configured for being placed within the imaging zone, wherein the radio frequency system is configured for suppling radio frequency power to the active B1 shim coil during the generation of the B1+ excitation field by the elliptical transmission coil, wherein the B1 shim coil is configured for shimming the B1+ excitation field within the imaging zone.

In a method of performing magnetic resonance imaging and a magnetic resonance apparatus, first MR data are acquired of a region of interest of a subject in the absence of a B1 field. Second MR data are acquired of the region of interest in the presence of a B1 field, and within a short time interval after generation of the B1 field. The first and second MR data are processed to determine a B1 field map, and a T1 map is generated using the B1 field map. The T1 map is a B1 corrected T1 map. The first and second MR data 103, 109 may be acquired as part of a T1 mapping sequence, such as a MOLLI or SASHA type cardiac T1 mapping sequence.

In a method and magnetic resonance apparatus for generating a B.sub.0 map of a region of interest, a magnetic resonance data set containing a number of image data sets is obtained and provided in a computer, wherein the image data sets are recorded using at least two measurement sequences and the mutually corresponding pixels of the image data sets each represent a time-dependent signal evolution. A B.sub.0 map of the region of interest is generated by the computer from the image data sets, wherein the B.sub.0 value of a pixel of the B.sub.0 map is determined from the associated signal evolution.

To calculate a high-resolution coil sensitivity distribution that does not depend on a shape or a structure of a subject with high accuracy. An MRI apparatus of the invention includes: a measurement unit that includes a reception coil including a plurality of channels, a measurement unit that measures a nuclear magnetic resonance signal of a subject for every channel of the reception coil; and an image computation unit that creates an image of the subject by using a sensitivity distribution for every channel of the reception coil, and a channel image obtained from the nuclear magnetic resonance signal measured by the measurement unit for every channel. The image computation unit includes a sensitivity distribution calculation unit that calculates a sensitivity distribution on a k-space for every channel by using the channel images and a composite image obtained by combining the channel images.

In a magnetic resonance fingerprinting method and apparatus for improved determination of local parameter values of an examination object, in which at least two signal comparisons of acquired picture element time series are carried out with comparison signal curves for determination of parameter values. A further (subsequent) signal comparison takes into account results of a preceding signal comparison. This multi-stage determination of parameter values allows an increase of the spatial resolution and the precision with which the parameter values can be determined.

A method of spatially imaging a nuclear magnetic resonance (NMR)parameter whose measurement requires the acquisition of spatially localized NMR signals in a sample includes placing the sample in an MRI apparatus with a plurality of MRI detectors each having a spatial sensitivity map; and applying MRI sequences adjusted to be sensitive to the NMR parameter. At least one of the MRI sequences is adjusted so as to substantially fully sample an image k-space of the sample. The remainder of the MRI sequences is adjusted to under-sample the image k-space. The method further includes acquiring image k-space NMR signal datasets; estimating a sensitivity map of each of the MRI detectors using a strategy to suppress unfolding artefacts; and applying the estimated sensitivity maps to at least one of the image k-space NMR signal data sets to reconstruct a spatial image of NMR signals that are sensitive to the NMR parameter.

Methods and systems are provided for predicting B.sub.1.sup.+ field maps from magnetic resonance calibration images using deep neural networks. In an exemplary embodiment a method for magnetic resonance imaging comprises, acquiring a magnetic resonance (MR) calibration image of an anatomical region, mapping the MR calibration image to a transmit field map (B.sub.1.sup.+ field map) with a trained deep neural network, acquiring a diagnostic MR image of the anatomical region, and correcting inhomogeneities of a transmit field in the diagnostic MR image with the B.sub.1.sup.+ field map. Further, methods and systems are provided for collecting and processing training data, as well as utilizing the training data to train a deep learning network to predict B.sub.1.sup.+ field maps from MR calibration images.

Systems and methods are disclosed for a simultaneous 3D T.sub.1 and B.sub.1.sup.+ mapping technique based on VFA imaging using a reference region VFA (RR-VFA) approach to eliminate the need for a separate B.sub.1.sup.+ mapping scan while imaging the prostate. The RR-VFA method assumes the existence of a reference region that is distributed throughout the volume of interest and is well characterized by a known T.sub.1 relaxation time. In particular, fat is generally selected as the reference region due to its distribution in the body. B.sub.1.sup.+ inhomogeneity is estimated in the fat tissue and interpolated over the entire volume of interest, thus eliminating the need for an additional scan.