Methods, systems, and apparatus, including computer programs encoded on computer storage media, for harmonizing diffusion tensor images. One of the methods includes obtaining a diffusion tensor image; determining a set of RISH features for the diffusion tensor image; processing a model input generated from the set of RISH features using a machine learning model to generate a model output identifying an image transformation from a set of image transformations, wherein each image transformation in the set of image transformations corresponds to a respective different first MRI scanner and represents a transformation that, when applied to first diffusion tensor images captured by the first MRI scanner, harmonizes the first diffusion tensor images with second diffusion tensor images captured by a reference MRI scanner; and processing the diffusion tensor image using the identified image transformation to generate a harmonized diffusion tensor image.

A system and method is provided for magnetic resonance imaging (MRI) guided respiratory gated radiotherapy using a respiratory motion model. MRI-guided respiratory gating is performed with a continuously updated model that represents a patient's internal anatomy as a mathematical function of an external respiratory surrogate. The motion model may be built and updated by acquiring images of a tissue in a subject and measuring, using the images, a position of the tissue in the images to determine motion of the tissue. The surrogate respiratory signal is acquired contemporaneously with acquiring the images. Motion of the tissue and the surrogate respiratory signal are correlated to create the motion model for the subject and gating a radiotherapy system may then be based upon the motion model. A multi-planar model-based respiratory gating may also be performed by sequentially imaging a stack of adjacent slice positions.

An MRI phantom for calibrated anisotropic imaging includes a plurality of separate sheathed taxons or integral taxons sharing common taxon walls, wherein each taxon has an inner diameter of less than 5 microns. The taxons form taxon filaments that are combined to form taxon ribbons. The taxons may have an average inner diameter of less than 1 micron, specifically about 0.8 microns with a packing density of about 1,000,000 per square millimeter. The filaments may include structural features such as an outer frame and crossing support ribs and may further include a visible alignment feature that allows for verifying orientation of an individual filament. The taxons may be formed as taxon fibers manufactured using a bi or tri-component textile/polymer manufacturing process. An anisotropic homogeneity phantom may include frame members that support fiber tracks extending in orthogonal directions, wherein each fiber track is formed of taxons.

A method for determining a brain age, the method comprising the following: providing a brain age determining convolutional neural network (CNN) (200); training the CNN (200) to determine the brain age based on a plurality of sets of input data comprising magnetic resonance imaging (MRI) scans of a brain, the set comprising at least two types of MRI volumes, wherein the at least one type of brain tissue on the first type of the MRI volume is represented by a different contrast with respect to other tissues than on a second type of the MRI volume; and performing an inference process using the trained CNN (200) to determine the brain age based on the set of input data comprising magnetic resonance imaging (MRI) scans of a brain, the set comprising at least the two types of the MRI volumes as used for the training.

The present invention relates to an apparatus (10) for detection of peripheral nerve stimulation. The apparatus comprises an input unit (20); at least one sensor (30); a processing unit (40); and an output unit (50). The input unit is configured to provide the processing unit with information on individual scans of a scan sequence of a Magnetic Resonance Imaging “MRI” unit being used to perform medical imaging of a patient. The at least one sensor is configured to acquire sensor data of the patient undergoing the medical imaging. The processing unit is configured to determine a presence of peripheral nerve stimulation “PNS”, wherein the determination of the presence of PNS comprises utilization of the information on the individual scans of the scan sequence and the sensor data of the patient. The output unit is configured to output an indication that the presence of PNS has been determined.

A method for applying a diffusion-weighting gradient during acquisition of diffusion-weighted imaging signals from a selected portion of a nervous system of a subject. Planar diffusion-weighted spin-echo (DWSE) imaging signals and planar diffusion-weighted stimulated-echo (DWSTE) imaging signals can be obtained to provide a plurality of sets of imaging signals. At least one set of imaging signals includes DWSTE signals that are associated with a high-b-value. A signal difference between DWSE imaging signals and DWSTE imaging signals can be corrected based on respective sets of DWSE imaging signals and DWSTE imaging signals having b-values at or near zero.

Techniques are disclosed for capturing scan data of an examination object via a magnetic resonance system. The techniques include capturing a first set of a diffusion-weighted scan data by excitation and, in an acquisition phase, acquiring a first echo signal, wherein before the acquisition phase in a diffusion preparation phase, diffusion gradients are switched for diffusion encoding of the scan data, The techniques additionally include capturing a second set of non-diffusion-weighted scan data by excitation and, in an acquisition phase, acquiring a second echo signal, wherein before the acquisition phase, in a diffusion preparation phase, the same diffusion gradients are switched as are switched for diffusion encoding of the scan data of the first set of diffusion-weighted scan data, although they have no influence on the second echo signal. Diffusion-weighted and non-diffusion-weighted scan data is thereby captured, having identical disturbances caused by eddy currents induced by switched gradients.

The present invention relates to the locally resolved examination of objects by means of magnetic resonance (MR) and relates specifically to a less motion-artifact prone method for navigated multi-shot acquisition of diffusion-weighted image data using moment-nulled magnetic field gradients for diffusion encoding. The invention further relates to an apparatus for performing the method.

The disclosure provides a diffusion kurtosis imaging method, which includes acquiring scan image signals of a scanned object; fitting the scan image signals using an unconstrained optimization algorithm to obtain elements of a first diffusion tensor and elements of a first kurtosis tensor; determining at least one type of parameters of diffusion tensor imaging parameters or kurtosis tensor imaging parameters based on the elements of the first diffusion tensor and the elements of the first kurtosis tensor; and generating a parameter image based on the at least one type of parameters of diffusion tensor imaging parameters or kurtosis tensor imaging parameters.

Techniques for correcting gradient non-linearity bias in mean diffusivity measurements by MRI systems are shown and include minimal number of spatial correction terms to achieve sufficient error control using three orthogonal diffusion weighted imaging (DWI) gradients. The correction is based on rotation of system gradient nonlinearity tensor into a DWI gradient frame where spatial bias of b-matrix is described by its Euclidian norm. The techniques obviate time consuming multi-direction acquisition and noise-sensitive mathematical diagonalization of a full diffusion tensor for medium of arbitrary anisotropy.