chapter 10 lecture mark d. herbst, m.d., ph.d

55
Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D.

Upload: douglas-hall

Post on 18-Dec-2015

236 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Chapter 10 Lecture

Mark D. Herbst, M.D., Ph.D.

Page 2: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 3: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Image Detail and Noise

• Two image characteristics that reduce the visibility of anatomy/pathology– Blurring – level of detail – related to pixel size– Visual Noise – random bright and dark dots

• These two are related, because if you improve one, you worsen the other.

Page 4: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 5: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 6: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 7: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Other image characteristics also impact visibility of

anatomy/pathology

• Camouflage

• Irregular borders

• Low contrast

• Motion (blurring)

Page 8: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 9: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 10: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 11: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 12: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 13: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

In-Plane ResolutionIn-Plane Resolution

This is a photo that has been taken at 165x256 resolution

Page 14: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

In-Plane ResolutionIn-Plane Resolution

This is a photo that has been taken with 329x512 resolution

Page 15: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

In-Plane ResolutionIn-Plane Resolution

Original Resolution 720x1150

Page 16: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 17: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 18: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 19: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 20: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 21: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

What is the MR image?

• Really just a display of the signal intensities from the voxels in the patient as pixels. No detail is seen within a voxel (pixel).

Page 22: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 23: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Voxels, FOV, and Matrix size

• Pixel size = FOV / matrix size

• Voxel size = pixel size x slice thickness

• FOV = field of view – can be square or rectangular (i.e., 25 cm x 25 cm, or 25 cm x 20 cm)

• Matrix = how many pixels in the image, square or rectangular (i.e., 256 x 256, or 256 x 192)

Page 24: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 25: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 26: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 27: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Exercise

• What is the voxel size?– FOV = 25 cm, Matrix = 256 x 256, ST = 1 mm– FOV = 13 cm, Matrix = 256 x 128, ST = 1 mm– FOV = 50 cm x 25 cm, Matrix = 512 x 256, ST

= 5 mm

Page 28: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

How detailed must the image be?

• The goal is to select the optimal voxel size for the anatomy and pathology you want to see

Page 29: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 30: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 31: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Sources of Noise

• The patient’s body

• Electrical devices – radio, TV, lights, motors, etc.

Page 32: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Signal-to-Noise Ratio (S/N)

• Two ways to increase S/N– Increase signal– Decrease noise

Page 33: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

How to increase S/N

• Increase voxel size—this increases S– Done by increasing ST or increasing pixel

size

• Increase field strength– increases S because S increases with

increasing B0

– Also increases N, but not as much as S, so S/N increases

Page 34: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

How to increase S/N

• Adjust TR and TE – longer TR and shorter TE lead to increase S, and therefore, increased S/N

• Use better RF receive coils– Body coils are worst, small surface coils are

best– New multicoil designs increase S

Page 35: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

How to increase S/N

• Decrease receive Bandwidth (BW)

• BW = 1/(dwell time)

• Problem of increased chem shift artifact with decreased BW

• If you increase BW to decrease chem shift, you get more noise

Page 36: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

How to increase S/N

• Signal averagingThe number of signals averaged (NSA), or the

number of excitations (NEX) or the number of acquisitions (NACQ) all mean the same thing. It is like taking multiple pictures and adding them together. The noise does not increase as much as the signal.

S/N ~ sqrt (NEX), so 4 NEX doubles S/N

Page 37: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Signal-to-NoiseSignal-to-Noise

Page 38: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Signal-to-NoiseSignal-to-Noise

Page 39: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Signal-to-NoiseSignal-to-Noise

Page 40: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

Signal-to-NoiseSignal-to-Noise

Page 41: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 42: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 43: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 44: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 45: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 46: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D
Page 47: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

AVERAGING

Page 48: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

BANDWIDTH

LARGE (WIDE) SMALL (NARROW)

Page 49: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

BANDWIDTH = 1/(DWELL TIME)

Page 50: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

DWELL TIME

SHORT sample of echo LONG sample of echo

Page 51: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

PELVIS

• FOV 340• RFOV 65• Slice 7mm• Matrix 256• Scan Percent 60• NSA 4• Time 5:54

Page 52: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

KNEE

• FOV 160• RFOV 80• Matrix 256• Scan Percent 90• Turbo Spin Echo• Slice 4 mm• NSA 2• Knee Coil• Time 2:28

Page 53: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

WRIST

• Gradient Echo• 3D Volume Acq.• FOV 100• RFOV 80• Matrix 256• Scan Percent 80• Slice 2.4 mm• Time 3:52

Page 54: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D

PROTOCOL FACTORS AFFECTING

IMAGE DETAIL IMAGE NOISE

• Slice Thickness• Matrix Size• FOV

• Slice Thickness• Matrix Size• FOV• Coil Selection• Averaging• Bandwidth• TR and TE• Field Strength

Page 55: Chapter 10 Lecture Mark D. Herbst, M.D., Ph.D