Echo Planar Imaging-Avinesh Shrestha

Echo planar imaging (EPI) is the method of rapid magnetic resonance imaging (MRI), overcoming one of the significant disadvantage of MRI concerning with slow imaging time. However, EPI-MRI imaging comes with it's own unique imaging artifacts.

1. Echo P lanar Imaging

2. Outline Introduction K-Space Basic principle Types Artifacts Clinical applications 2

3. Introduction Echo-planar imaging (EPI) is capable of significantly shortening magnetic resonance (MR) imaging times. Echo-planar imaging allows acquisition of images in 20–100 msec. This time resolution virtually eliminates motion-related artifacts. Therefore, imaging of rapidly changing physiologic processes becomes possible. Echo-planar images with resolution and contrast similar to those of conventional MR images can be obtained by using mult-ishot acquisitions in only a few seconds. 3

4. K space Before discussing the basic principles of echo planar imaging or any other MR imaging technique, it is important to present the concept of k space. k-space is the matrix containing the raw data from all the measurement before the Fourier Transformation. Rectangular in shape and has two axes perpendicular to each other ( kx and ky ). Each number in k-space refers to a spatial frequency. MR images are obtained by performing FT of k-space. 4 The wavenumber ( k ) is the number of waves or cycles per unit distance .

5. T scan = NSA x N PE x TR 5

6. K space The individual points ( kx,ky ) in k-space do not correspond one-to-one with individual pixels ( x,y ) in the image. Each k-space point contains spatial frequency and phase information about every pixel in the final image. Conversely, each pixel in the image maps to every point in k-space 6

7. K space Facts about k-space- The center of k-space (i.e., low spatial frequencies) contains information on large-scale structures (e.g., contrast between large objects). The periphery of k-space (i.e., high spatial frequencies) contains information on the fine structures (e.g., edges and small-scale details). Positive polarity phase fills the top half of K space; negative polarity fills the bottom half Steep gradients, (both positive and negative), select the most outer lines and shallow gradients select the center lines Provided no phase errors occur during data collection, k-space possesses a peculiar mirrored property known as conjugate (or  Hermitian )  symmetry.Data are symmetrical in K space. 7

8. 8 As an overview, the center of  k -space contains low spatial frequency information, determining overall image contrast, brightness, and general shapes. The periphery of  k -space contains high spatial frequency information (edges, details, sharp transitions).

9. 9 As an overview, the center of  k -space contains low spatial frequency information, determining overall image contrast, brightness, and general shapes. The periphery of  k -space contains high spatial frequency information (edges, details, sharp transitions).

10. K space 10 Phase-conjugate symmetry. About half of k-space is sampled by reducing the number of phase-encoding steps. The other half of k-space is synthesized/reconstructed .

11. K space 11 Although phase-conjugate symmetry reduces imaging time while preserving spatial resolution, this is accomplished at the expense of the signal-to-noise ratio (SNR).  For half-Fourier imaging, SNR is reduced by a factor of √½ or 30% less than a comparable sequence using the full number of phase-encoding steps . In theory, phase-conjugate symmetry allows one to acquire data using only half the normal number of phase-encoding steps, thus potentially reducing imaging time by as much as 50%. In practice, the time savings is closer to 40%, but this is still a huge benefit widely used in modern MRI protocols. 

12. K space trajectories The main k space trajectories are: Standard rectilinear/Non-EPI trajectory EPI trajectory Radial trajectory Spiral trajectory 12 Rectilinear

13. Standard trajectories Sequential Centric Reverse Centric View-Sharing Keyhole FSE/TSE Hal Fourier/Partial Fourier Zero padding 13

14. 14

15. 15

16. Types of non EPI trajectories Sequential: Acquire each line in separate readout in order from top to bottom. Used in standard imaging Centric: Acquire each ky -line in a separate readout, starting at the center line of k-space ( ky = 0) and alternating up and down on successive excitations. For applications in which intent is to have initial relative contrast dominate the image contrast such as after a preparatory RF pulse or contrast agent injection. 16

17. Types of non EPI trajectories Keyhole trajectory Acquire a complete k-space matrix, one line at a time, typically in order from top to bottom of k-space. Then, periodically update the central lines, reconstructing them with the older peripheral lines. Used for imaging dynamic events in which producing multiple images with high temporal resolution is more important than highly accurate spatial resolution. Interpolation: Acquire the central ky (at least half the desired matrix size, ) and replace the missing data with zeros. This “zero padding” in k space is equivalent to interpolating a low resolution image to make a higher resolution image. applied for situations in which reducing the scan time by a factor of Nacq / Ndesired is worth the associated loss of SNR in proportion to the square root of that same factor. 17

18. Types of EPI trajectory EP trajectories sweep back and forth across k-space very rapidly such that the entire k-space matrix, or a large portion of it, is filled in one execution of the pulse sequence. If the entire k-space matrix is filled in one execution of the pulse sequence, then it is termed a single-shot acquisition . If instead more than one execution of the pulse sequence is required to collect all the desired k-space data , then the scan is a multi-shot acquisition and scan time is TR times the number of shots (executions). 18

19. Types of EPI trajectory Single-shot: Traverse all of k-space, covering a rectilinear grid, starting in one corner, sweeping across the kx axis for a given ky value, then jumping (“blipping”) to the next ky value and sweeping back across the kx axis. Another blip to the next ky value follows and so on until all of k-space is traversed. For applications requiring very fast acquisitions. Especially used for functional MRI (fMRI) studies due to high speed and sensitivity to BOLD effect. Distortions and T2* dependent signal loss typically prohibit the acquisition of a high resolution image. 19

20. Types of EPI trajectory Multishot -EPI: Sparsely traverse all of k-space, starting in one corner, sweeping across the kx axis for a given ky value, then jumping (“blipping”) to a different ky value and sweeping back across the kx axis. Another blip to another ky value follows and so on until a fraction of k-space equal to 1/(#shots) is filled. For applications requiring very fast acquisitions but not as fast as single shot EPI. May also be used to fill a larger k-space matrix and thus obtain better spatial resolution than possible with single shot EPI. Since time spent acquiring data is shorter than with single shot EPI, distortions and signal loss are less severe 20

21. Basic principle I n echo-planar imaging, multiple lines of imaging data are acquired after a single RF excitation. Like a conventional SE sequence, an SE echo planar imaging sequence begins with 90° and 180° RF pulses. However, after the 180° RF pulse, the frequency-encoding gradient oscillates rapidly from a positive to a negative amplitude, forming a train of gradient echoes . Each echo is phase encoded differently by phase-encoding blips on the phase-encoding axis. Each oscillation of the frequency-encoding gradient corresponds to one line of imaging data in k space, and each blip corresponds to a transition from one line to the next in k space. This technique is called blipped echo-planar imaging 21

22. 22 Conventional SE imaging. Within each TR period, the pulse sequence is executed and one line of imaging data or one phase-encoding step is collected

23. Echo-planar imaging. Within each TR period, multiple lines of imaging data are collected 23

24. EPI data acquisition As with conventional MR- imaging, we put the acquired data for the frequency and phase encoding into the 2D grid called k-space. Also, the 2D Fourier transform is used to create the image. In EPI, the data is filled into k-space in a rectangular “zig-zag”-like pattern.

25. Hardware The requirements for gradient strength, rise time, and duty cycle are markedly increased because all of k space should be traversed in a single RF excitation by using a rapidly oscillating frequency-encoding gradient Echo-planar imaging is accomplished by using gradient coils capable of a maximum amplitude of 20 mT /m, a minimum rise time of 0.1 msec , a slew rate of 200 T/m per second, and a duty cycle of 50%–60% Higher bandwidth ADC in the order of MHz, upto 2 MHz ( in conventional sequence- kHz)

26. Unique features of EPI SNR Contrast Resolution 26

27. SNR The SNR in EPI is inherently low due to wide bandwidth. For same parameter ,SNR is lower in EPI than in conventional MRI. Can be increased by Use of surface coils ,multi-shot imaging 27

28. Contrast in EPI Different contrasts are achieved by either beginning the sequence by variable RF excitation pulse in GE– EPI or with 90° and 180° RF pulses in SE– EPI with any type of RF pulse. EPI– FLAIR (180° / 90° / 180° followed by EPI readout) nulls the CSF but the sequence is significantly faster than in conventional FLAIR sequencing. Proton density or T2 weighting is achieved by selecting either a short or long effective TE. T1 weighting is possible by applying an inverting pulse before the excitation pulse to produce saturation. Contrast in SE EPI is similar to that of CSE. GRE EPI provides T2* weighting. 28

29. Resolution Resolution depends on the maximum gradient amplitude-time product in the raw data. To increase resolution requires: increased gradient amplitude which may ultimately result in unacceptable safety problems) increase in gradient duration (lowers the effective image bandwidth and increases the sensitivity of the images to shape distortion and other artifacts) or both. Higher bandwidth ADC in the order of MHz, upto 2 MHz ( in conventional sequence- kHz) 29

30. Resolution EPI resolution is limited by the SNR of the images. Since typical EPI data are collected over 40 to 50 msec , as compared to the nearly 1 second of acquisition time spent on a conventional scan, the SNR is down by a factor of more than 4-fold at comparable resolution on this basis alone. With EPI at 3.0 T , spatial resolution can be improved tremendously up to 0.75 X 0.75 mm for a 2.5-mm slice thickness; thus susceptibility artifacts at the bone-air interface do not constitute a problem. At the present time, it seems that EPI spatial resolution is largely gradient limited 30

31. Types of epi Types of EPI: Single shot EPI Multishot EPI Preparation of EPI Spin echo EPI Gradient echo EPI Hybrid EPI Inversion recovery EPI DW EPI 3D EPI 31

32. Single shot EPI Collects all the data required to fill lines of k-space from a single echo train. Multiple echoes are generated, and each is phase encoded by a different slope of gradient to fill all the required lines of k-space in a single-shot. Rapid change in gradient polarity rephases the FID produced after the RF excitation pulse to generate gradient-echoes. As the frequency-encoding gradient switches its polarity so rapidly, it is said to oscillate. The phase-encoding gradient also switches on and off rapidly, but its polarity does not need to change in this type of k-space traversal. No TR in single-shot imaging .TR is therefore infinity. 32

33. Multishot EPI The SS-EPI’s main advantage is reduced scan-time, it suffers from several limitations including low signal-to-noise ratio (SNR), vulnerability to image distortions (or artifacts), and limited attainable spatial-resolution due to the fast signal decay . In order to overcome these limitations, multi-shot EPI (ms-EPI) can be used. Multi-shot EPI follows the same principles as ss -EPI except it uses multiple excitations to acquire k-space data. One way to acquire the data is in an interleaved manner, where the acquired lines of each shot fill in the gaps from the other shots. As an example, assuming 64 lines of k-space were desired, in two-shot EPI only 32 lines would be acquired for each excitation. In an interleaved acquisition the first shot would acquire lines 1,3,5,7, etc. while the second shot would acquire lines 2,4,6,8, etc. 33

34. MS- EPI Each shot traverses the y-direction of k-space much faster than in ss -EPI, and this results in reduced distortions that occur in ss -EPI due to its slow ky sampling rate. One of ms-EPI’s main advantages is its ability to acquire a much greater number of k-space lines. In ss -EPI the total number of attainable lines is limited because of the T2∗ signal decay of a single excitation. In ms-EPI the number of lines in each shot is also limited by T2∗ decay, but by combining data from each shot the total amount of lines is increased, and is limited only by the number of excitations used. This results in a larger k-space coverage and therefore improved spatial resolution In ms-EPI the number of k-space lines acquired in each shot is not stretched to the maximum as in ss -EPI, the acquired signal from each shot does not decay as much as much as in ss -EPI, resulting in better SNR. The number of shots commonly used in practice are currently two, four, and eight, with two/four-shot EPI being the most typical. 34

35. SE-EPI 35

36. GE- EPI 36

37. Hybrid EPI 37

38. IR-EPI 38

39. DWI -EPI 39 Diffusion weighted imaging. Two diffusion gradient are place on either side of 180 refocusing pulse. These gradients have certain b value which depend the strength the duration of the gradient and the time between two gradient lobes .the echo generated are readout in epi

40. VOLUME-EPI Idea of Echo Volume Imaging (EVI) is a direct extension of EPI to 3D imaging using a single RF excitation. Incorporates the oscillatory EP readout along the frequency-encoding and phase-encoding axes. For the slice-select phase encoding a constant gradient is applied during the complete readout (or small blips) so that the complete 3D data is encoded. 40

41. 41

42. EPI ARTIFACTS There are four major artifacts in EPI: T2*weighting signal loss Ghosting Geometric distortion & Susceptibility artifacts Fat water chemical shift 42

43. Geometric distortion Geometric distortions manifest themselves as pixel movements predominantly in the phase encoding direction in EPI images. Appearance of stretching or bunching of signal near in the vicinity of field inhomogeneity. The spatial distortions are greatly significant in the phase-encoding direction. The pixels are therefore shifted in the direction in the phase encoding gradient 43

44. 44

45. 45

46. How to reduce distortion? Attempts to reduce the total acquisition time of EPI so the result is less affected by magnetic disturbance and generate images that have less artifact: Partial Fourier Reconstruction Parallel imaging Segmented EPI Multi-shot 46

47. Partial fourier Partial echo: Is performed when only part of echo is read during application of frequency encoding gradient. The peak of echo is usually centered in the middle of sampling window. When a very short TE is required, the echo rephases sooner than with a long TE. This technique switches on the frequency encoding gradient as soon as it is possible to do so, but also moves the peak of echo so that it is no longer at the centred in the middle of the sampling window and occurs sooner. This means that only the peak and dephasing part of echo are sampled and therefore initially only half of the frequency area of k space is filled (right side of k space) 47

48. Partial fourier Partial echo: However due to right –to-left symmetry of k space ,the system can extrapolate the data in the right side and place it in the left side. Therefore although initially only the right side of the k space is filled with data, after extrapolation both sides contain data and overall no data are lost. Partial echo imaging is routinely used when a very short TE is selected in the scan protocol. 48

49. Partial fourier Partial averaging: The negative and positive lobes of k-space on each side of the phase axis are symmetrical and a mirror image of ach other. As long as over the half of the lines are filled during the acquisition ,the system has enough data to create an image . Scan time decreases as a result distortion also decreases. Partial averaging is used where a reduction in scan time is necessary and where the resultant signal loss is not of paramount importance. 49

50. 50

51. 51

52. Parallel imaging Distortion in EPI is driven by the temporal separation between subsequent odd and even echoes within the EPI readout train. One way to reduce distortion in EPI is, therefore, through the use of parallel imaging with either generalized auto calibrating partial parallel acquisition (GRAPPA) or sensitivity-encoding (SENSE) . GRAPPA and SENSE speed up the traversal of k-space 52

53. 53

54. How to reduce distortion? however, image distortion in parallel imaging–enhanced EPI DWI remains problematic particularly at high field strengths and for high-resolution imaging Another way to reduce distortion in DWI is to speed k-space traversal through the use of alternate trajectories. These methods include but are not limited to: Multishot EPI sequences interleaved spiral sequences and fast spin-echo (FSE)–based sequences such as periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER) short-axis PROPELLER EPI readout-segmented EPI 54

55. 55

56. 56

57. Propeller EPI Traverses k-space radially in groups parallel and centered about the k(0,0) point. The groups, also known as “blades” are rotated about the k(0,0)point of k-space like unto a rotating propeller. Advantages include a more rapid coverage of the blade in k-space for a scan duration similar to EPI speeds and a high coverage of the outer portions of k-space. The technique maintains the robustness of radial sampling to bulk motion of the object but without the associated blurring due to reduced higher k-space acquisition. 57

58. 58

59. 59

60. 60

61. 61

62. T2* signal loss The transverse magnetization is T2* dependent. The effect of field inhomogeneities is dependent upon the size of the phase dispersion caused relative to the size of the pixel. Intra-voxel dephasing is where field inhomogeneities occur within a voxel, resulting in signal loss due to phase dispersion. More prevalent in GE-EPI, as the 180 degree pulse in SE-EPI acts to reverse phase dispersions and reduce dependence on T2*. 62

63. 63

64. Chemical shift Arises from the inherent 3.5 ppm frequency difference between fat and water protons under the influence of an external magnetic field. Chemical shift artifact causes the misplacement of signal from fat in the image which is a misregistration of fat and water protons from a voxel that are mapped to different pixels. It can also create both signal voids and signal superimposition (high signal) in areas where fat and water interface. Because the readout BW is so high, in EPI, chemical shift is not manifested in frequency readout direction but in phase encoding direction if present. 64

65. 65

66. 66

67. 67

68. 68

69. How to reduce chemical shift? Apply saturation band/Fat saturation Decrease echo spacing Under sample 69

70. N/2 GHOSTING Images appear displaced by half the field-of-view (FOV) in the phase encoding direction, with the bottom half of the (real) image generating a ghost image that appears at the top of the FOV and the top half of the (real) image generating a ghost image that appears at the bottom of the FOV. The readout gradients for odd and even echoes are of different polarity (one positive, the other, negative).  While in MRI, any magnetic disturbance (such as inhomogeneity, eddy current, susceptibility, gradient imperfection) will cause the center of the echo to shift and results in the mismatch of centers of odd and even echoes.  It is due to this mismatch that the N/2 ghost appears. 70

71. 71 Simplified representation of signal collected in: a) conventional sequences, b) EPI. One can notice that center of echoes from odd and even echoes appear to be misaligned in EPI . A typical N/2 ghost manifestation

72. 72

73. 73

74. 74

75. Suspectibility artifacts predominantly due to the modality's high sensitivity to inhomogeneities in the magnetic field. Extremely detrimental to the image when caused by metal implants causing irrecoverable signal. To a lesser extent, found at air-tissue boundaries and are less detrimental In the brain, found near the ear cavities and near the sinuses. 75

76. Clinical applications Brain imaging Abdominal imaging Cardiac imaging miscellaneous 76

77. Brain imaging Brain Imaging EPI–based diffusion imaging is routinely used for evaluation of early cerebral ischemia and stroke EPI–based perfusion imaging and DWI EPI is performed to evaluate cerebral ischemia and differentiate different types of tumor from radiation necrosis Evaluation of cortical activation with echo-planar imaging or functional MR imaging is an active area of research in neuroscience 77

78. 78 Large right middle cerebral artery infarct in a 61-year-old woman. The infarct was seen on a cerebral angiogram; MR imaging was performed approximately 5 hours after onset of the stroke. Axial T1- weighted (a) and T2-weighted (b) images show only mild sulcal effacement and gyral edema on the right side of the brain (arrow). Axial diffusion-weighted (c) and perfusion (d) images show the extent of the infarct clearly

79. Abdominal imaging The primary contribution of echo-planar imaging to abdominal imaging has been the increased gradient capabilities of echo-planar imaging the result has been shorter TEs and improved GRE sequences for breath-hold imaging and three-dimensional MR angiography. There are several potential uses for echo-planar imaging in abdominal imaging. Multishot echo-planar imaging can be used to acquire breath-hold T2- weighted images, potentially replacing SE and fast SE T2-weighted sequences. In the uncooperative patient, single-shot echo-planar imaging of the upper abdomen can be performed in about 2 seconds. 79

80. Abdominal imaging Single-shot techniques, such as half-Fourier acquisition single-shot fast SE (HASTE) imaging or single-shot fast SE (SSFSE) imaging, can be used for breath-hold imaging but are limited by image blurring and suboptimal contrast resolution. the echo-planar imaging pulse sequence can potentially combine the tissue contrast advantages of conventional SE sequences and the speed of single-shot fast SE sequences 80

81. 81 Colon cancer metastatic to the liver. (a) Axial fast SE image shows one faint lesion (arrow). (b) Axial single-shot fast SE image also shows only one lesion (arrow). (c) Axial echo-planar image shows many lesions (arrows).

82. Cine Imaging Echo-planar imaging facilitates rapid evaluation of cardiac function and anatomy. Cine imaging of the heart is performed with GRE echo-planar imaging over multiple cardiac cycles. When singleshot echo-planar imaging is used, electrocardiographic gating is not required. Therefore, this technique is particularly helpful in patients with arrhythmias. 82