MRI IMAGE QUALITY and Factors affecting it by T.R.B.

All the aspects of image quality are discussed with proper images and physics behind that.

1. MRI Image Quality Thumba Raj Baruwal B.Sc. MIT 2 nd Year Roll No.162

2. Contents SNR CNR Spatial Resolution Scan Time References.

3. Introduction Image quality is a basic concept that applies to all types of images including photographic and video images as well as a wide variety of images produced for medical purposes. At the most fundamental level, image quality is a comparison of the image to the actual object. In MRI, image quality is directly related to its usefulness in providing an accurate diagnosis. The usefulness of an image can only be assessed on a case-by-case basis. However, the true test of the quality of a specific image is whether it serves the purpose for which it was acquired.

4. Signal to Noise ratio (SNR) The SNR is defined as the ratio of the amplitude of signal received to the average amplitude of the background noise. Signal is the voltage induced in the receiver coil by the precession of coherent magnetization in the transverse plane at time TE. Noise represents frequencies that exist randomly in space and time. Signal is cumulative and predictable. It occurs at, or near, time TE and at specific frequencies at, or near, the Larmor frequency. Noise, on the other hand, is not predictable and is detected by the whole coil volume . It occurs at all frequencies and is also random in time and space.

5. Signal to Noise ratio (SNR) The main source of noise is from thermal motion in the patient but it is also generated by background electrical noise of the system. Noise is constant for every patient and depends on the build of the patient, the area under examination, and the inherent noise of the system. The purpose of optimizing SNR is to make the contribution from signal larger than that from noise.

6. Signal to Noise ratio (SNR) Factor that affects signal amplitude affects the SNR. These are as follows: Voxel volume Magnetic field strength of the system Proton density of the area under examination Coil type and position TR, TE, and flip angle Number of signal averages (NSA) Receive bandwidth

7. Voxel Volume The building unit of a digital image is a pixel. The brightness of the pixel represents the strength of the MRI signal generated by a unit volume of patient tissue (voxel). Voxel dimensions are determined by the pixel area and the slice thickness . The pixel area is determined by the field of view (FOV) and image matrix. Large voxels contain more spins or nuclei than small voxels and therefore have more nuclei to contribute toward signal. Large voxels consequently have a higher SNR than small voxels.

8. Voxel Volume

9. Voxel Volume

10. FOV FOV is given in cm or mm depending on the vendor and defines the length of MR image coverage in each plane. The FOV can be given as a single number such as 24 cm for square FOV or as 24 × 20 cm for rectangular FOV. The rectangular FOV is chosen mainly to reduce the scan time and the smaller FOV almost always refers to the FOV in phase encoding direction. The FOV is directly proportional to the SNR of the resulting image.

11. FOV If we double the FOV in frequency direction, then image SNR will be 2 times better as well. If we halve FOV in both frequency and phase direction from 30 to 15 cm, then SNR will be 4 times less. This remarkable reduction in SNR can also be confirmed visually with the more grainy or noisy image with smaller FOV.

12. FOV Sample images acquired with 30 cm ( a ) and 15 cm ( b ) FOV are shown on the same volunteer.

13. FOV Sagittal brain using FOV of 120mm Sagittal brain using a square FOV of 240mm

14. Image Matrix MR imaging matrix do not have any measurement units but it merely shows the number of how many sample points or measurements we are making during the acquisition. If you see 512 × 512 imaging matrix, you can tell that this image was acquired with 512 points in frequency and 512 points in phase direction. Usually the lower imaging matrix will be applied in phase direction to reduce the total scan time. The effect of imaging matrix on resulting image is inversely proportional to square root of the matrix factor.

15. Image Matrix

16. Image Matrix Sample images acquired with 128 × 128 matrix ( a ) and 320 × 320 matrix ( b ) FOV are shown on the same volunteer.

17. Changing the Image Matrix Sagittal brain using 256 phase matrix Sagittal brain using 128 phase matrix

18. Slice Thickness It is defined in mm and determines the depth of voxel on slice encoding direction. For 3D imaging, the slice thickness is usually much thinner and can vary from 0.5 to 5 mm depending on the application and coverage of anatomy. The thicker the slice is, the higher the SNR becomes.

19. Slice Thickness Sample images acquired with slice thickness of 3 mm ( a ) and 15 mm ( b ) are shown on the same volunteer.

20. Slice Thickness

21. Magnetic field strength As field strength increases, so does the energy gap between high- and low-energy nuclei increases . As the energy gap increases, fewer nuclei have enough energy to align their magnetic moments in opposition to B0. Therefore, the number of spin-up nuclei increases relative to the number of spin down nuclei. The NMV increases at higher field strengths, and there is more available magnetization to image the patient.

22. Magnetic field strength

23. Proton density The number of protons in the area under examination determines the amplitude of received signal. Areas of low proton density in terms of those that are MR-active such as the lungs have low signal and therefore low SNR, whereas areas with a high proton density such as the pelvis have high signal and therefore high SNR. The proton density is inherent to the tissue and cannot be changed that is why it is an intrinsic contrast parameter. When measures are taken to null or saturate signal from a tissue, SNR decreases because signal contribution from that tissue is removed.

24. Proton Density

25. Types of coil The types of coil affects the amount of received signal and therefore the SNR. Larger coils receive more noise in proportion to signal than smaller coils because noise is received from the entire receiving volume of the coil. Quadrature coils increase SNR because several coils are used to receive signal. Phased array coils increase SNR as the data from several coils are added together. Surface coils placed close to the area under examination also increase SNR.

26. Types of coil In general, the size of the receiver coil should be chosen such that the volume of tissue optimally fills the sensitive volume of the coil. Large coils, however, increase the likelihood of aliasing, because tissue outside the FOV is more likely to produce signal . The position of the coil is also very important for maximizing SNR. To induce maximum signal, the coil must be positioned in the transverse plane perpendicular to B0. Angling the coil, as sometimes happens when using surface coils, results in a reduction of SNR .

27. Surface Coils Surface coils are the simplest of the coil designs. They are either a circular or a rectangular loop of wire. Surface coils are placed very close to area under examination. Generally used for spines , shoulder ,neck ,TMJ etc. Signal can be obtained from depths which are (50-75)% of the diameter of the coil. They have very good SNR.

28. Surface Coils

29. Volume Coils Volume coils need to have the area of examination inside the coil. They can be used for transmit and receive but sometimes they are used for receive only. Head coils have very good depth coverage but low SNR. The higher the number of transmitter and receiver elements , the better the SNR ,foe example a 32 channel (receiver element) body coil will produce better SNR compared to a 4 channel body coil. Volume coils can be of: Paired saddle coil. Helmholtz pair coil. Birdcage coil.

30. Volume Coils

31. Coil position and SNR

32. TR,TE ,and Flip angle Although TR, TE, and flip angle are usually considered parameters that influence image contrast they also affect SNR and therefore image quality.

33. TR and SNR The TR controls the amount of longitudinal magnetization that recovers before the next RF excitation pulse is applied. A long TR allows full recovery of longitudinal magnetization so that more is available to be flipped into the transverse plane in the next TR. A short TR does not allow full recovery of longitudinal magnetization so less is available to be flipped. The SNR improves as the TR increases. Increasing the TR also increases scan time and the chance of patient movement.

34. TR

35. TR and SNR Changing TR at 3T

36. Flip Angle and SNR The flip angle controls the amount of transverse magnetization created by the RF excitation pulse, which induces a signal in the receiver coil. If the TR is long, maximum signal amplitude is created with flip angles of 90° because full recovery of longitudinal magnetization occurs with a long TR, and this is fully converted into transverse magnetization by a 90° flip angle SNR significantly decreases in the lower flip angle image. If the TR is short, the flip angle required to generate maximum transverse magnetization and therefore signal is less than 90° and is governed by the Ernst angle equation .

37. Flip Angle in GRE and SNR Greater the flip angle ,the more T1 weighting . The longer the TE ,the more T2* weighting.

38. Flip Angle and SNR

39. TE and SNR The TE controls the amount of coherent transverse magnetization that decays before an echo is collected. A long TE allows considerable decay of coherent transverse magnetization before the echo is collected, while a short TE does not. SNR decreases as the TE increases because there is less transverse magnetization available to rephase and produce an echo. This explains why T2-weighted sequences, which use a long TE, usually have a lower SNR than T1 or PD-weighted sequences that use a short TE.

40. TE and SNR

41. TE and SNR Changing TE at 3T

42. Number of Signal Averages (NSA or NEX) NSA is an indication of how many signal averages taken during an MR acquisition. This is usually achieved by repeating the acquisition in frequency direction and taking an average of the sampled signal. NSA is also called number of excitations (NEX). NEX or NSA is mainly used to increase SNR at the expense of increased acquisition time. SNR increase is directly proportional to the square root of NEX increase factor.

43. Number of Signal Averages (NSA or NEX) SNR is doubled by simply increasing the NEX fourfold. But fourfold increase in NEX also means a fourfold increase in total scan time.

44. Number of Signal Averages (NSA or NEX)

45. Number of Signal Averages (NSA or NEX)

46. Sampling During the sampling window, the system samples or measures spatial frequencies in the echo. Every sample results in a data point that is placed in a line of k-space. Each data point contains information about spatial frequencies in the echo at different time points during the sampling window. The rate at which sampling occurs is called the digital sampling rate or digital sampling frequency . The digital sampling frequency determines the time interval between each data point. This time interval is called the sampling interval and is calculated by dividing the digital sampling frequency by 1

47. Sampling If the digital sampling frequency is 32 000 Hz (32 KHz ), the sampling interval is 0.031 ms (1 ÷ 32 000). If the digital sampling frequency halves to 16 000 Hz (16 KHz ), the sampling interval doubles to 0.062 ms (1 ÷ 16 00)

48. Sampling The most optimum digital sampling frequency is determined by the Nyquist theorem. (The full name is Whittaker– Kotelnikov –Shannon–Raabe–Someya–Nyquist theorem). The Nyquist theorem states that when digitizing a modulation of several frequencies, the highest frequency present in the modulation must be sampled at least twice as frequently to accurately digitize or represent it.

49. Sampling Sampling at less than once per cycle represents a completely incorrect frequency that leads to an artifact called aliasing. High digital sampling frequencies also result in more noise data.

50. Receive Bandwidth This is the range of frequencies that are accurately sampled during the sampling window . Receive bandwidth = K (Gradient field * FOV). The receiver bandwidth is the reciprocal of the total sampling time. The unit of the receiver bandwidth is hertz(Hz)/pixel. Reducing the receive bandwidth results in less noise sampled relative to signal.

51. Receive Bandwidth SNR vs receive bandwidth

52. SNR relationship

53. Contrast-to-Noise ratio (CNR) CNR is defined as the difference in the SNR between two adjacent areas. Image contrast depends on both intrinsic and extrinsic parameters. Intrinsic contrast parameters are those that cannot be changed because they are inherent to the body’s tissues . Intrinsic contrast parameters are as follows: T1 recovery time T2 decay time Proton density (PD) Flow Apparent diffusion coefficient (ADC)

54. Contrast-to-Noise ratio(CNR) Extrinsic contrast parameters are those that can be changed because they are under our control . The extrinsic contrast parameters are as follows : TR TE Flip angle TI Turbo factor/echo train length

55. Contrast-to-Noise ratio(CNR)

56. Administration of Contrast Gadolinium based contrast agents or simply gadolinium contrast agents are molecular complexes containing the rare earth gadolinium ,chelated to a carrier ligand. GBCAs are a type of paramagnetic contrast agents , which are the primary class of MRI contrast media. The gadolinium molecules shorten the spin lattice relaxation time. As a result, on T1 weighted images they have a brighter signal. GBCAs have a number of uses like detection of focal lesions (tumors, abscess , metastasis) , MRA , calculating MR perfusion parameters.

57. Administration of Contrast Axial T1 weighted image of the liver without (left) and with (right) manganese contrast . The enhanced image shows enhancement of normal liver so that the liver pathology is darker

58. T2 weighting A T2 weighted image contrast predominantly depends on the differences in the T2 decay times between fat and water. T2 weighting is specifically used to increase the CNR between normal and abnormal tissue. Pathology is often bright on a T2 weighted image as it contains water. However the SNR, spatial resolution and scan time are usually compromised in these images because of the parameters selected.

59. T2 weighting

60. T2 weighting

61. T1 Contrast Short TR and short TE helps to generate T1 contrast.

62. Echo Train Length or Turbo Factor ETL or turbo factor can be defined as the number of refocusing 180° rf pulses after the initial excitation rf pulse. FSE or TSE sequences with higher number of echoes are discovered after the initial spin echo pulses and reduced the MR image acquisition time significantly for T2 and PD weighted images. The higher ETL or turbo factor usually means a higher T2 weighting, shorter acquisition time, less number of slices, more motion artifact, and more blurry images.

63. Echo Train Length or Turbo Factor

64. Magnetization transfer contrast (MTC). Magnetization transfer imaging (MTI) is a technique by which radiofrequency (RF) energy is applied exclusively to the bound pool using specially designed MT pulses. Some energy that is used for saturation of bound pool protons is then transferred to the free water pool primarily via dipole dipole interactions and free water pool becomes partially saturated . If the free water pool is subsequently imaged using routine RF pulses and gradients , its signal will be reduced secondary to the MT effect.

65. Magnetization transfer contrast (MTC) The magnitude of this MT effect can be quantified by obtaining two sets of images (one with an MT pulse and one without it) and then digitally subtracting them. MTR = (So − SMT)/So The relative difference in signal between two adjacent tissues is known as magnetization transfer contrast (MTC).

66. TOF MRA without (left) and with (right) MT suppression. Better visualization of small vessels after MT pulse.

67. Flow-related techniques There are techniques that are specifically designed to produce signal only from nuclei with certain characteristics. Nuclei that do not possess such characteristics do not produce signal, and so there is a good CNR between them and those that do. For example, phase-contrast MR angiography is a technique that only images flowing spins. Stationary spins produce no signal, and so there is a good CNR between vessels and the tissue around them.

68. Flow-related techniques

69. Chemical Suppression Techniques These techniques are used to suppress signal from either fat or water or other substances such as silicone. They improve the CNR because the signal from unwanted tissue is removed. They are often used in conjunction with T2 weighting so that pathology has a high signal compared to the unwanted tissue. Fat is the commonest tissue to suppress. There are a variety of ways to do this based on the fact that fat and water either precess at different frequencies or have a different relaxation time .

70. Fat Suppression in MRI

71. Inversion Recovery Very good SNR as the TR is long. Excellent T1 contrast. In IR sequence SNR can decrease as tissue are suppressed.

72. STIR

73. FLAIR Axial T2 weighted FLAIR of the brain

74. Spatial Resolution Spatial resolution is the ability to distinguish between two points as separate and distinct, and is controlled by the voxel size. Small voxels result in high spatial resolution, as small structures are easily differentiated. Large voxels, on the other hand, result in low spatial resolution, as small structures are not resolved so well. In large voxels, individual signal intensities are averaged together and not represented as distinct within the voxel. This is called partial voluming .

75. Spatial Resolution Spatial resolution is entirely controlled by the size of the voxel. The voxel is defined as the pixel area multiplied by the slice thickness. Therefore the factors that affect the voxel volume are: slice thickness FOV image matrix

76. Pixel size Pixel size is determined by the distance traveled in k -space. Pixel size in the phase axis of the image is determined by the steepest phase-encoding gradient, both positively and negatively. Pixel size in the frequency axis of the image is determined by the sampling window.

77. Voxel volume Large voxels have higher signal than small ones because there are more spins in a large voxel to contribute to the signal. Any setting of FOV, image matrix size or slice thickness that results in large voxels leads to a higher SNR per voxel. As the voxels increase in size, resolution decreases. There is thus a direct conflict between SNR and resolution in the geometry of the voxel.

78. Changing the matrix Changing the image matrix alters the number of pixels that fit into the FOV. As the image matrix increases, pixel and thus voxel size decrease, assuming the FOV is unchanged. This increases resolution but reduces SNR. Changing the phase matrix also changes scan time.

79. Changing the matrix and resolution

80. Changing the FOV As reducing the FOV affects the size of the pixel along both axes, the voxel volume is significantly reduced. Decreasing the FOV therefore has a drastic effect on SNR. Using a small FOV is appropriate when using small coils that boost local SNR.

81. Changing slice thickness and resolution Changing the slice thickness changes the voxel volume proportionally and results in a change in both SNR and resolution. 10 mm slice thickness 3 mm slice thickness

82. Scan Time As multiple slices are selected during two- and three-dimensional volumetric acquisitions, movement during these types of scan affects all the slices. To optimize the scan time :select Short TR Low phase matrix Low NSA High turbo factor in TSE rectangular FOV

83. Improving Resolution without increasing scan time Changing the frequency matrix only. The frequency matrix does not affect scan time but, if increased, increases resolution. Using asymmetric FOV. This maintains the size of the FOV along the frequency axis but reduces the FOV in the phase direction. Therefore, the resolution of a square FOV is maintained, but the scan time decreases in proportion to the reduction in the size of the FOV in the phase direction. This option is useful when anatomy fits into a rectangle

84. K space The outer lines of K space contain data with high spatial resolution, as they are filled by steep phase‐encoding gradient slopes. The central lines of K space contain data with low spatial resolution , as they are filled by shallow phase‐encoding gradient slopes. The central portion of K space contains data that has high signal amplitude and low spatial resolution. The outer portion of K space contains data that has high spatial resolution and low signal amplitude.

85. K space and Image quality Image using central K space data points only

86. K space and Image quality K space signal and resolution data

87. Summary Increasing the matrix size: Increases spatial resolution Decreases signal Increases scan time Increasing NEX Increases signal Increased scan time Less noise Fewer artifact due to signal averaging. Increasing the FOV Increases the signal Lower spatial resolution Increased viewing area Increasing the slice thickness Increases the signal Decreases the resolution Increases the partial volume effect

88. Summary To improve the SNR: Increase NEX Lower resolution Thicker slices Larger FOV Use surface coils To improve the resolution: Increase the matrix Decrease the FOV Decrease the slice thickness

89. Summary

90. References MRI in practice 5th edition Catherine Westbrook . MRI Handbook_ MR Physics, Patient Positioning, and Protocols. Radiology Tutorials YouTube channel. Class Notes

91. THANK YOU