Picker International, Inc.

Koninklijke Philips Corporation N.v.

Mark J Loncar

Rao P Gullapalli

Haiying Liu

Kecheng Liu

Michael R Thompson

James A Halamek

Heidi A Schlitt

Gordon D DeMeester

Ulrich Wolfgang Katscher

Wayne R Dannels

Paul M Margosian

James B Murdoch

Yansun Xu

Neil Palmer

Thomas P Orefice

A magnetic resonance imaging system acquires a final image of a selected field of view with a selected spatial resolution. A magnetic resonance imaging scanner () encodes and receives magnetic resonance samples in phase encode and readout directions using a plurality of receive coils (). The encoding and receiving undersamples in the readout direction. A reconstruction processor () reconstructs magnetic resonance samples acquired by each of the plurality of receive coils () into a corresponding plurality of intermediate reconstructed images. Each intermediate reconstructed image has aliasing and in some aspects degraded high spatial frequency characteristics due to the reduced sampling in the readout direction. A combining processor () combines the plurality of intermediate reconstructed images based on coil sensitivity factors () to produce the final reconstructed image with the selected field of view and the selected spatial resolution in the readout direction.

Mr imaging with sensitivity encoding in the readout direction

A method of magnetic resonance imaging includes subjecting a region of a patient being imaged to magnetic resonance imaging pulse sequences thereby generating magnetic resonance echoes emanating from the region. The echoes are encoded with magnetic gradients and collected as k-space data in separate consecutive acquisitions. Each acquisition successively fills in one of a number of distinct subsets of k-space based on the encoding of the collected echoes for that acquisition. In turn, a reconstruction algorithm is applied to the k-space data after each acquisition to generate a series of temporally updated image representations of the region. Further, a magnetic resonance imaging apparatus for performing the method is also provided.

Shifting sampling window and interleaves sparse K-space data acquisition for improved temporal resolution

The present invention is directed to an MRI apparatus including a main magnet (,) that generates a substantially uniform temporally constant main magnetic field, B,, through an examination region (,) wherein an object being imaged is positioned. A magnetic gradient generator produces magnetic gradients in the main magnetic field across the examination region (,). A transmission system includes an RF transmitter (,) that drives an RF coil (,) which is proximate to the examination region (,). A sequence control (,) manipulates the magnetic gradient generator and the transmission system to produce an MRI pulse sequence. The MRI pulse sequence induces an initial set of contiguous spin echos and a subsequent set of gradient echos which stem from the object. A reception system includes a receiver (,) that receives and demodulates the spin and gradient echos. Ultimately, a reconstruction processor (,) reconstructs an image of the object from data collected via the reception system, and an output device produces a human viewable rendering of the image.

Spin and field echo (safe) FSE

A method of calibrating pre-emphasis gradient pulse conditioning for long term eddy current compensation in an MRI system is provided. It includes positioning an object in the MRI system for imaging therewith. A long term eddy current generating pre-scan gradient pulse is applied, and after a delay, an EPI sequence is run such that a resulting image of the object is acquired. This process of applying the pre-scan gradient pulse followed by the EPI sequence is repeated while varying the delay therebetween. Size variations in the resulting images of the object are observed, and these size variations are fit to an exponential curve to obtain one or more time constants for the long term eddy currents. Next, a reference EPI sequence is run without applying the long term eddy current generating pre-scan gradient pulse such that a reference image of the object is acquired. The long term eddy current generating pre-scan gradient pulse is reintroduced, and after a delay which is shorter than the time constants, a test EPI sequence is run such that a test image of the object is acquired. The process of applying the pre-scan gradient pulse followed by the test EPI sequence is repeated while adjusting the pre-emphasis gradient pulse conditioning which is based upon the time constants obtained. This continues until the test images of the object have a size substantially equivalent to that of the reference image.

EPI image based long term eddy current pre-emphasis calibration

After exciting magnetic resonance (60), a train of magnetic resonance echoes including echoes (70.sub.1, 70.sub.2, . . . ) is induced, e.g., with a series of refocusing pulses including pulses (72.sub.1, 72.sub.2, . . . ). The echoes are phase and frequency-encoded appropriately for generating at least first and second images (94, 98). The first image echoes are interleaved with the second image echoes, for example, odd-numbered echoes for the first image and even-numbered echoes for the second image. To select the effective first/second image echo time, the first/second image echo which follows resonance excitation by a time closest to the selected first/second image effective echo time is given the zero or minimum phase-encoding. Nearby echoes being encoded with low phase-encode angles. In this manner, the effective echo times of the first and second images are selected to have preselected relative contrasts of diagnostic interest. Typically, the effective echo times of both the first and second image is less than the time between resonance excitation and the eighth echo, even although the echo train may be 32, 64, 128, or more echoes.

Dual contrast fast spin echo with alternating phase-encode

Resonance is excited in a first slab (12) and manipulated to generate a plurality of data lines (16, 18) which span a fraction of k-space, e.g. a quarter of the phase encoding steps along a y-direction. Resonance is then excited in a second slab (22) displaced from the first slab and another series of data lines are generated. A resonance is excited and data lines generated in a plurality of additional slabs (32, 42). A resonance is excited in a slab (52) which partially overlaps the slab (12), e.g., has three of four slices in common. A series of data lines in the slab (52) are phase encoded with a different fraction of k-space. Two sets of differently phase encoded data sets have been generated in the example of FIGS. 2a and 2b. This process is cyclically repeated exciting in slabs which partially overlap proceedingly generated slabs, each time generating a different fraction of the data lines until a completed set of data is generated in an extended volume larger than one slab, slices 4-16 in the example of FIGS. 2a and 2b.

Multi-volume slicing and interleaved phase-encoding acquisition for 3 D

A method (100) of automatically tuning a radio frequency transmitter (24) and receiver (38) in a magnetic resonance imaging apparatus to an optimum frequency includes generating and acquiring (104) a magnetic resonance signal. The magnetic resonance signal is transformed to the frequency domain and spectral magnitude of the signal is computed (106). A center of gravity interpolation is performed (110) on the spectral magnitude to obtain a desired frequency sampling. A model function is generated based upon a strength of a main magnetic field which has peaks separated by the same separation as that for fat and water signals. The spectral magnitude is correlated (112) with the model function and a peak having the greatest magnitude is located therefrom. The location of a species peak along the spectral magnitude is estimated (114) from empirically derived information, the strength of the main magnetic field and the location of the correlation peak. A region of the spectral magnitude to be examined is defined (116) from empirically derived information, and a peak within this region is located (118) which is nearest the estimated location. Ultimately, the frequency of the radio frequency transmitter and receiver is set (122) to align with the peak located within the region that is nearest the estimated location.

Automatic frequency tuning for MR scanners

A plurality of radio frequency excitation pulses or shots (52) are applied, ten in the embodiment illustrated in FIGS. 4 and 5. Following each shot, sets of data lines are collected. In the first set, an early gradient echo (EGE1), a spin echo (SE1), and a late gradient echo (LGE1), are induced to form three corresponding data lines. Magnetization is inverted (56) and a second set of data lines are generated. In the illustrated embodiment, nine sets of data lines are generated in each repetition. Phase-encoding gradient pulses (86, 88) are applied to cause the early gradient echo, the spin echo, and the late gradient echo data lines of each set to fall offset by a third of k-space. Phase-encoding pulses (74) are applied before each set and stepped such that in half of the repetitions, the phase-encoding increases with each subsequent set. In the other half of the repetitions, the phase-encoding decreases for each subsequent set. In this manner, the first and last data line in each segment of k-space are from the same echo position with the repetition. By selecting the intermediate phase-encoding step with which to start the first set, the phase-encoding at the center of k-space is selectively adjustable such that the pseudo echo time is selectively adjustable.

Ultra-fast imaging technique using k-space segmentation with minimum

A sequence controller (40) controls the pulses applied by the radio transmitter (24) and the gradient amplifiers (20) and gradient coils (22) such that each repetition includes a prepreparation sequence segment, such as a presaturation sequence segment and a magnetization transfer contrast correction (MTC) segment, and a plurality of image sequence segments. More specifically, each of the image sequence segments induce resonance, phase and frequency-encode the resonance, and generate one or more views of data, all within a corresponding one of a plurality of slabs or sub-regions (74.sub.1, 74.sub.2, . . .) of an image volume (72). More precisely to the preferred embodiment, the imaging sequence segments interleave the slabs such that resonance is not excited concurrently in adjacent slabs, without exciting resonance and collecting a view in a non-adjacent slab. The views are sorted (80) by slab and stored in corresponding slab data memories (82). The data in each slab data memory is reconstructed (84) into a corresponding slab image stored in corresponding slab image memories (86). The slab images are stacked or combined (88) into a volumetric image representation of the region of interest and stored in a volume image memory (90).

Batch multi-volume angiography using magnetic resonance imaging

A subject's respiratory cycle is monitored (70) and analyzed (74) to determine a probability table (76). The probability table is continuously updated to provide a dynamic indication of the center between most stationary and most moving halves or other characteristic points within the respiratory cycle. At the beginning of each repeat time (TR), as resonance is excited (40), the output of the respiratory monitor is checked (80) and a determination is made (82) whether the patient is currently in the most stationary half or the most moving half of the respiratory cycle. Data lines generated in one half of the respiratory cycle are phase-encoded (84, 88) to generate a data line in each of a plurality of segments of k-space on one side of the central, zero phase-encoding point. Echoes occurring during the other half of the respiratory cycle are phase-encoded to generate data lines in segments of k-space on an opposite side of the central phase-encode angle. This process is repeated until a complete data set is generated with data lines generated during the stationary half of the respiratory cycle on one side of k-space and data lines generated during the half of the respiratory cycle with more motion on the other side. These data lines are then reconstructed into an image representation.

Respiratory motion compensation using segmented k-space magnetic

A sequence control (40) causes a transmitter (24) and gradient amplifiers (20) to transmit appropriate radio frequency excitation and other pulses to induce magnetic resonance in selected dipoles and cause the magnetic resonance to be refocused into a series of echoes following each excitation. A receiver (38) converts each echo into a digital data line. Each data line is regridded (70) for uniformity in k-space (FIG. 4). The data lines are one-dimensionally Fourier transformed (72) in a frequency encode direction. The one-dimensionally Fourier transformed data lines are multiplied (80) with a phase correction vector. A phase correction vector determining system (82) determines a corresponding phase correction vector for each echo number or position following excitation from a series of calibration echoes. To compensate for a decrease in magnitude with echo position following excitation, the intensity of each data line is scaled (90) to a common magnitude. The intensity corrected data lines are stored in a memory (96) and one-dimensionally Fourier transformed in a phase encode direction into an image memory (100). A monitor (104) converts the image memory representation into a human-readable display.

Magnetic resonance scan calibration method for ultra-fast image

In a magnetic resonance imaging system, a read gradient scaler (102) scales the amplitude and width of read gradients (78, 82, 84). A sampling rate control (104) controls the sampling rate of the resonance signals received from corresponding magnetic resonance echoes (80 54, 86). For example, when the amplitude of the gradient pulse is doubled and its width halved, the sampling rate of the resultant magnetic resonance signal is doubled, e.g., from a bandwidth of 32 MHz to a bandwidth of 64 MHz. In this manner, some echoes are read-out over a longer period of time with a lower bandwidth to produce lower signal-to-noise data lines; whereas, other echoes are much shorter and are read-out more quickly, but with a lower signal-to-noise ratio. The phase-encoding gradients (74, 88, 90) are controlled (70) such that the resultant data lines with the lowest bandwidth have the lowest phase-encoding angles closest to the zero phase-encoding and the data lines with the highest bandwidth have the highest phase-encoding angles closest to the negative and positive maximum phase-encoding gradients. The data lines are sorted (110) by phase encode angle and bandwidth and reconstructed (120) into an image representation.

Ultra-fast MR imaging data acquisition scheme using mixed bandwidth data

Magnetic resonance is excited (50) in first and second species dipoles of a subject in a temporally constant magnetic field. The resonance is refocused (52) to generate a spin echo (54) centered at a time when the first and second species resonance signals are in-phase. Gradients echoes (64, 68) are generated, centered at a time (2n+1).pi./.delta..omega. before and after the spin echo, where .delta..omega. is a difference between the first and second species resonance frequencies. In this manner, the first and second species signals are 180.degree. out-of-phase in the gradient echoes. The resonance is refocused (82) one or more times to generate additional spin and gradient echoes with different phase encodings (78). The sequence is repeated with yet more phase encodings, and magnetic resonance signals from the spin echo and the two gradient echoes are reconstructed (86) into a spin echo image (s.sub.0) and a pair of gradient echo images (s.sub.+1, s.sub.-1). A phase map is generated (90) from the spin and gradient echo images. One of the gradient echo images is corrected (116) with the phase map. The phase corrected gradient image is additively combined (118) with the spin echo image to generate a first species image (112) and is subtractively combined (120) to generate a second species image (114).

Three point technique using spin and gradient echoes for water/fat

Magnets (12) create a temporally constant magnetic field through an examination region (14). Radio frequency coils (26, 34) and a transmitter (24) transmit radio frequency saturation pulses (52) and the resonance excitation and manipulation pulses of a magnetic resonance imaging sequence (72) into the examination region. Gradient amplifiers (20) and gradient coils (22, 32) create magnetic field gradients across the examination region for spatially focusing the saturation, for spoiling (62, 66, 70) residual transverse magnetization and for frequency and phase encoding in the magnetic resonance imaging sequence. A sequence controller (40) includes a saturation pulse controller (44) for generating the saturation pulse (52) and slice select gradients (58) and a steady state sequence controller (48) for generating the imaging sequence (72). The saturation is spectrally focused by limiting the frequency of the saturation pulse to selected frequencies. The saturation is spatially focused by applying the slice or other spatially limiting gradient pulses concurrently with the radio frequency saturation pulse. A phase shift controller (76) increments the phase of the RF saturation pulse by a preselected phase increment, e.g., 105.degree.. Preferably, the phase shift increment increases linearly in each cyclic repetition, i.e., increases 105.degree. in the first repetition, 210.degree. in the next repetition, 315.degree. in the next repetition, etc. A reconstruction processor (80) reconstructs resonance signals generated during the imaging sequences into an image representation for storage in a memory (82) and display on a video monitor (86).

Growing community of inventors

Richmond Heights, OH, United States of America

Mark J Loncar