Echo-planar imaging (EPI)

is a data acquisition strategy used in MR imaging permitting very rapid data acquisition. The method was originally described by Mansfield in 1977 and employs the following imaging strategy. Rather than acquiring a single image line (in k space) after the preparation phase of the pulse sequence, the entire MR image is acquired. Multiple variations of this image acquisition strategy have been devised since its inception, but the basic concept is that multiple rather than single image lines are acquired after spin preparation.

Any pulse sequence can be divided into a spin preparation module, during which the spins are disturbed in their alignment, and a readout module during which the signals coming from the realigning spins are detected by the MR imager (Fig.1). This preparation module identifies the type and hence name of a sequence (spin echo pulse sequence, gradient echo pulse sequence, inversion recovery IR pulse sequence) and is used also in EPI. Hence, the names gradient-echo echo-planar, or spin-echo echo-planar. The special feature of an EPI sequence is that many rather than single lines are read during the readout module after a single preparation module using gradient rephasing. This imposes important additional perfomance requirements on an MR imager compared to a standard MR system because many image lines have to be read in a very short time. As can be seen in Fig. 1, the readout of multiple imaging lines occurs under a free induction decay FID in gradient-echo EPI with the signal envelope given by a decaying exponential, which decays with the effective spin - spin T2 relaxation time. The same is true for spin-echo EPI, where the signal envelope of the echo increases and then decreases exponentially with the time constant T2*. T2* is short and in some organs like the heart can have values in the range of 20-30 ms, while in the brain it is typically around 100 ms at 1.5 Tesla. This limits useful measurement time. Measurement time is determined by the data collection time for a single line, which in turn is dependent on the spatial resolution required, the strength of the applied gradient fields and the time the system takes to ramp the gradients between lines. In an MR imager with maximum gradient field strengths of 5-10 mT/m, only a few image lines with adequate signal can be read. Readout of further lines is suppressed by the decaying exponential (Fig. 1).

Successful echo-planar imaging requires that the time taken to read a single image line (TSL) be much smaller than T2* and that many lines can be read before the exponential has decayed to approximately half its value. This necessitates strong gradient fields with rapid switching capability. The latter can be achieved in two ways. The original design used so-called oscillating gradient system. In such systems an oscillating circuit consisting of the inductor of the gradient coil and a capacitor drives the gradient. The very fast gradient systems now available on clinical MR systems can be ramped linearly as on older systems but at much higher gradient field strengths and slew rates. They are often referred to as nonoscillating or nonresonant gradient systems. With these gradient systems any gr susceptibility artefacts

Therefore, various modifications of the original echo-planar pulse sequence have been proposed which lead to substantial improvements in image quality. Dividing the image lines (in k-space) into several blocks or interleaves (Fig.2), preceded by a preparation pulse, results in drastic improvements in image quality (Fig.3). As an example, rather than collecting 128 lines after a single alpha pulse in a gradient-echo echo-planar sequence, four interleaves of 32 lines are collected. This reduces TR and TE roughly by a factor of 4, thus reducing artefacts, and also partly removes the problems associated with the T2* decay under which image lines are sampled. In gradient-echo echo-planar imaging the time penalty for introducing a few additional alpha pulses is minimal (Fig.4), and thus interleaved echo-planar imaging is the method of choice when high resolution echo-planar imaging of organs with short T2* is performed. In spiral scanning, the image is scanned in a spiral fashion and interleaving of spirals is also possible. Scanning then occurs by moving along intertwined spiral trajectories which all originate at the centre of k-space. Interleaved scanning therefore permits the operator to choose the appropriate sequence from a continuum of sequences spanning from standard to echo-planar sequences. Blocks of lines can be acquired sequentially or if cardiac gating is used, spread over the appropriate number of heartbeats. If echo-planar readout is used in interleaves within a RARE pulse sequence, the resulting sequence is called a GRASE pulse sequence.

GvS

GvS