Fetal fMRI

1
Fetal fMRI
Nottingham Lace

• Penny Gowland

Magnetic Resonance Centre and School of Human
Development University of Nottingham
2
Summary

• How did we come to do this?
• Why are we doing this?
• Description of studies

Sagittal reconstruction breech fetus 0309
3
Why did we do this?

• To study fetal brain development in fetal growth
  restriction (FGR)
• Cerebral palsy affects more than 1500
• 10 of CP arises during labour
• Current measures of fetal brain function are
  indirect
• Prenatal study of fetal brain function

T2 weighted EPI
4
Brain growth in FGR

• Ultrasound finds apparent brain sparing in FGR
• MR does not

5
Aim of fetal fMRI project

• To determine nature of detectable fetal brain
  activity (e.g. Mum v Dad).
• To determine the sequence of development in the
  magnitude and extent of activation in the fetal
  brain.
• To study the effect of fetal compromise on brain
  functional and anatomical development.

6
How did we come to do this?

• We have a long term interest in fetal imaging

Perfusion rate ml/100g/min
7
How did we come to do this?

• We have a long term interest in Echo-planar
  imaging
• Multiplanar image formation using NMR Mansfield
  J. Phys. C., 10, L55-58, 1977
• Therefore we have possibly the best whole body
  EPI scanner in the world
• No Niquist ghosts
• Unfortunately it can ONLY do EPI
• Unfortunately it is nearly 20 years old

8

How did we come to do this?
36 week fetus
Whole body EPI
Fetal Brain
Placenta
Fetal Lung
Fetal Liver

• Susceptiblity not really a problem
• Fetus matched to surrounding
• Operating at 0.5T- low CNR but quiet
• Except ?? iron in maternal bowel

9
How did we come to do this?

• We have a long term interest in fMRI

Change in separation of centre of activation in
post central gyrus p0.00048 Normals(6) 11 ?
2 mm Dystonics (5) 4.4 ? 0.9 mm
Little Finger
Index Finger
Both Fingers
Bowtell and Francis, Nottingham
10
How did we come to do this?

• We had a medic doing a PhD in physics.

11
Studies so far

• Pilot study female voice reading a nursery story
  
• Auditory study, Spanish guitar music (HBM).
  Related study of MCA pulsatility index and FHR
  changes.
• Vibroacoustic stimulator
• Visual study

12
General Methods

• 0.5 T purpose built scanner
• NRPB guidelines
• Informed written consent from all subjects
• T2 weighted EPI
• effective echo time 65 ms (minimum used)
• 0.5 kHz switching frequency
• Initial localization
• Ultrasound
• Multislice EPI (9 mm slice thickness)

13

Pilot study Paradigm

• Mother recorded a nursery rhyme (music used for
  adults)

14
Pilot study Results

• 1/4 fetuses moved too much for analysis
• 2/4 fetuse showed statistically significant
  activation p 0.017 (5.3 ) p0.0001(2.3 )
• 1/4 fetuses showed no activation (p0.25)
• 2/2 adults showed actiavtion in auditory cortex
  (plt0.001), with signal changes of 1 and 3

15
Auditory study Subjects

• 20 subjects
• 16 subjects
• 4 controls (blinded for person doing the
  analysis)
• Music played through an MR compatible headphone
  attached to maternal abdomen, 85 dBSPL at surface
• For controls music was played to mother through
  same headphones at 70 dBSPL
• 15 s on/15 s off, for 30 cycles

16
Auditory study Stimulus
Paradigm

• Fetal heart rate monitored during study
• TEAM monitor, modified to be MR compatible

17

Auditory study MRI

• Images acquired at 2 Hz (each slice sampled at
  1/3 Hz), low rate increases audibility of
  auditory stimulus
• 30 saturation volumes collected initially
• EPI
• 128x128
• Te 70ms
• 128x128 matrix, in plane resolution 5x5 mm,
  (4x4 mm for adults), slice thickness 15mm, to
  increase dephasing, 6 slices

18
Auditory study Motion correction
Motion correction

• Gross motion is detected using
• A MOTION PLOT (mean signal intensity of 3 pixels
  in high contrast region of the image versus time)
• FHR monitor
• If motion was apparent in a cycle in both plots,
  the cycle was removed from further analysis

19
Auditory study Motion correction
time (minutes)
Motion plot Mean signal in ROI including a
boundary of fetal brain
20

Auditory study Motion correction

• Remaining cycles averaged, giving a high S/N
  image of the mean position of the brain
• Used to construct an object map in Analyze, to
  mask the region of the images to be used in image
  registration.
• Images viewed in movie mode, with the object map
  overlaid. Cycles in which the fetus moved outside
  the object map were also removed.

21

Auditory study Motion correction

• Prior to image registration, maternal signals
  must be removed, as mother and fetus will move
  independently, therefore remaining images masked
  with object map, to remove maternal data.
• Images globally normalised, and interpolated to
  cubic voxels.
• Registration performed using AIR within Medx.

22

Auditory study Motion correction

• Checked visually (susceptiblity artefacts makes
  this fail)
• Spatial filtering with 2D Gaussian kernel of 4.26
  mm fwhm
• Low pass temporal filtering with Gaussian kernel
  of 1.4 s fwhm
• High pass temporal filtering by zeroing lowest 6
  frequency Fourier components

23
Auditory study Statistical analysis
Auditory study

• Correlation with activation paradigm, convolved
  with a Poisson function (TTP 6 s, taken from
  adult data)
• Gaussian Random Fields theory used to give a
  corrected SPM

24
Auditory study Fetal results
Auditory study Adult results

• 2 mothers moved too much (back pain)
• 3 subjects showed susceptiblity artefact
• 2 subjects showed no activation (p0.4 in TL)
• 5 subjects showed activation (at plt0.009)
• 2 LTL, 1 RTL, 1 both TL, 1 frontal, also 2 SS
• Mean signal change was 2.2 ? 0.3

25
Auditory study Control results
Auditory study

• 1 control could not be analyzed (susceptibilty)
• 3 controls showed no activation

Eyes
Nasal cavity
26
Auditory study Adult results
Auditory study

• 2 adults no acitvation
• 3 adults showed significant activation
• 2 left only, 1 both sides
• signal change 1.1 ? 0.14

27
Auditory study Fetal time course
28
Averaged Time course
Auditory study Fetal time course
29
Auditory study Cardiovascular

• 20 normal pregnancies at term
• Randomisation to study/ control group
• Stimulus Spanish guitar music 90 dBSPL
• Pre-stimulus
• Baseline MCA Doppler pulsatility index
• computerised FHR recording
• During stimulus FHR monitored
• Post-stimulus MCA Doppler (blind)
• Non-parametric tests for statistical analysis

30
Auditory study Cardiovascular
Change in middle cerebral artery pulsatility index
31
Auditory study Cardiovascular

• Middle cerebral artery pulsatility index (MCA-PI)
  was significantly lower post-stimulus than
  pre-stimulus in the study group (p0.007)
• Control group showed no significant change
  post-stimulus (p0.44)
• The change in MCA PI (pre-post) was significantly
  different in the two groups (p0.01)

32
Auditory study Cardiovascular
Cardiovascular study
160
Stimulus
Pre
Post
150
140
Pre
130
Post
Control
120
10
10
10
10
N
No significant change in FHR during stimulus
33
Auditory study Conclusion

• We have mapped fetal auditory cortex in two
  separate studies
• signal changes quite large
• fetal and adult haemoglobin have different
  affinities for oxygen. The susceptiblity of fetal
  haemoglobin has not been measured.
• Scanner very noisy (100 dBSPL)

34
Auditory study Conclusion

• Using US we have found a related increase in MCA
  blood flow, which we assume is related to
  increased neuronal activity
• This may provide a simple measure of fetal brain
  function
• We have demonstrated that this decrease in MCA PI
  was independent of FHR

35
Visual Study Motivation

• Previous fetal heart rate response to constant
  intensity light source has been reported

36
Visual Study Paradigm

• Light source Kingbright 50 LED cluster (2.5 W)

37
Visual Study Paradigm

• 30 cm light pipe (cardboard tube lined with
  aluminised plastic)
• This avoids stimulus correlated field distortions
  due to current switching, and also susceptibility
  effects
• No measurable effect on image of phantom

38
Visual Study Paradigm
Paradigm

• 1100-1200 Lux at maternal abdomen
• Constant amplitude it could not be flashed
• Sufficient to pass through 5 cm chicken flesh
• Intensity compared to a white endoscopy light

39
Visual Study Subjects
Subjects

• 12 subjects
• 9 study subjects
• 3 controls (light pipe blocked off)
• 36 weeks gestation (head engaged)
• Preliminary ultrasound scan used to determine
  position of eyes with respect to abdomen
• posterior facing fetuses excluded

40
Visual Study MRI
MR methods

• 10 slices, 10 mm slice thickness
• 128x128 matrix, 4x4 mm in-plane resolution
• Effective echo time 110 ms (T2 matched)
• TR 4 s per volume
• 45 volumes acquired prior to paradigm
• 40 cycles acquired
• 8 s light on,
• 16 s light off

41
Visual Study Results

• 3 controls no activation (pgt0.95 in all cases)
• 1/9 not analyzed due to motion
• 3/9 no activation at plt0.01 level
• 5/9 showed activation
• p 0.00004, 0.001, 0.00001, 0.0001, 0.0085
• 4 frontal, 1 temporal (p0.0085)
• No activation in V1

42
Visual Study Results
1330
1463
1462
1483
No activation in primary visual area, but
consistent frontal activation
43
Visual Study Results
44
Visual Study Conclusion

• It was difficult to localize precisely frontal
  activation within fetal brains, due to lack of
  landmarks on T2 weighted fetal brains
• Future work will include acquiring a HASTE image
  for localization
• Need a flashing visual stimulus

45
Visual Study Conclusion

• No V1 activation
• light not flashed- no adult activation either
• light intensity might be modulated by blood flow
• light intensity may be too low
• V1 is known to mature slowly anyway
• Frontal activation
• not consistently reported in adults
• much more stimulating stimulus to a fetus in a
  dark environment

46
Fetal HDR- Methods

• Event related (short) stimulus used to monitor
  fetal haemodynamic response
• 4 s on 16 s off
• Vibro-acoustic stimulus used
• expect some sensory and acoustic responses
• Commercial VAS (Corometrics Model 146) modified
  in-house to be MR compatible

• Similar sound played through MR compatible
  headphones to adults

47
Fetal HDR- results

• Results obtained for 7/15 fetuses and 5/13 adults
• Fetuses
• Time to peak 7.4 0.9 s
• mean change of 2.7 0.9 .
• Adults
• Time to peak 6.5 0.5 s
• mean change of 1.0 0.4 .
• Response slightly later and stronger in fetuses

48
Fetal HDR- analysis

• The time courses were fitted to a gamma variate
  function

49
Fetal HDR- results
Fetal Adult
50
Fetal HDR- conclusions

• The fetal HDR is similar to that of the adults.
• For an event related paradigm it may be slightly
  delayed, but this effect is only small
• It does seem to be larger than expected
• Measure susceptibility of fetal haemoglobin

51
Improvements to Methodology

• For cycles selected for analysis, the motion plot
  to be correlated to stimulus time-course
• Investigate intake of iron supplements
• Improve anatomical definition of scans (T2
  weighting and spatial resolution)
• SPARSE imaging - problem with motion

52
Future Experiments

• Find a paradigm that will elicit a differential
  response (e.g. Mum v Dad?)
• This will offer an extra dimension to study in
  fetal development and fetal compromise.
• We assume that an auditory stimulus will be most
  useful, but
• differential attenuation by maternal abdomen
• evolving frequency dependence of auditory
  pathways during pregnancy

53
Summary

• Studies of fetal brain response to auditory and
  visual stimulation
• Methods of dealing with motion artefacts have
  been developed
• However we continue to scan near term, which is
  very hard work for the research team, so more
  serious NMR methods of motion tracking are being
  investigated

54
Conclusion

• MRI has the potential to become an important tool
  in studying normal and abnormal fetal brain
  development
• Activation is detected in about 50 of subjects
  (including technical problems)
• This relevant to conditions such as FGR leading
  to subtle or serious brain damage
• It may also be relevant for understanding adult
  brain function

55
Acknowledgements

• Obstetricians
• Ian Johnson
• Phil Baker
• Keith Duncan
• David James
• Bryony Strachen
• Eileen Bradley
• Shantilla Vadeyar

• Physicists
• Jon Fulford
• Rachel Moore
• Damien Tyler
• Sue Francis
• Jon Hykin
• Sir Peter Mansfield
• Richard Bowtell
• Paul Clark
• Ron Coxon

Our volunteers
56
(No Transcript)
57
SPARSE-like results

• Problems probably due to interference of the BBC,
  or change in paradigm?
• alternating stimulus
• 1 no registration (SPARSE) (not analyzed)
• 1 baby moved too much (not analyzed)
• original stimulus
• 1 susceptiblity artefact (not analyzed)
• 1 no activation
• 2 to be analyzed

58
Possible SPARSE paradigms
Cycle 24s
SPARSE
SEMI-SPARSE
59
Improvements to Methodology- SPARSE

• Hall et al
• Too few scans to allow registration, will try an
  IR-EPI sequence to improve contrast for image
  registration
• Quasi sparse, with varying flip angles to
  maintain constant signal considered
• Semi-sparse, with two volumes scanned per cycle
  now being attempted in a few subjects ? No
  habituation to scanner.

60
MRI of fetal brain development

• Cerebral development by MRI follows (lags) that
  found neuroanatomically
• Abnormal brain development reported in fetuses
  with congenital CNS abnormalities
• Postnatal MRI has shown impaired cerebral
  development after FGR

T2 weighted EPI
Levine Radiology, 201(3), 715-758, 1999
61
Fetal brain maturation by MRI

• The NMR relaxation parameter T2 is related to
  tissue structure and fat content
• Hypothesize that it will change in the fetal
  brain during gestation.
• Change may be delayed in compromised pregnancies
• Rate of maturation and myelination in utero v
  preterm may be different

62
Fetal brain maturation by MRI

• Recruitment from antenatal clinics
• 35 Normal subjects 18-40 weeks GA
• 6 Compromised subjects 30-40 weeks GA
• Compromised defined as IUGR (IBR lt 3rd percentile)

63
Fetal brain maturation by MRI
Frontal lobe
64
Fetal brain maturation by MRI
65
What is BOLD?

• Increased neuronal activity causes increased
  blood flow, blood volume concentration of oxyHb
• Oxyhaemoglobin is relatively diamagnetic
• increased concentration on activation
• Deoxyhaemoglobin is relatively paramagnetic
• decreased concentration on activation

66
What is fMRI?

• These changes in magnetic state of blood, and
  blood volume, lead to changes in spin dephasing
  and very small changes in MR signal
• BOLD
• Blood oxygenation level dependent effect
• Haemodyanmic response time lag (6 sec.)

67
What is fMRI?
Detecting activation
M.R. SIGNAL UNACTIVATED pixel
time
68
Pilot study SPM
Results
69

Auditory study Motion correction
Eyes
Temporal lobe
70
Auditory study Fetal SPM
71
Fetal movement plot
Auditory study Motion correction
Mean signal in ROI including a boundary of fetal
brain
72
Auditory study T2 results

• T2 of fetal brain ? 190 ms
• T2 of adult brain ? 90 ms
• This has implications for the choice of echo time
  in fetal fMRI studies (optimally TE T2 )
• Differences could be due to
• lack of air in fetal sinuses
• lack of myelin in fetal brain
• differences between fetal and adult haemoglobin

73
Auditory study Conclusion

• We have mapped fetal auditory cortex in two
  separate studies
• signal changes quite large
• fetal and adult haemoglobin have different
  affinities for oxygen. The susceptiblity of fetal
  haemoglobin has not been measured.
• Baseline T2 long
• Scanner very noisy (100 dBSPL)

74

Future Experiments

• take mothers voice, process it to retain gross
  amplitude and bandwidth whilst altering vocal
  characteristics (IHR- QS)
• determine the acoustical features of a sound that
  will evoke a differential response in the fetus
  in terms of
• position of activation
• area of activation
• change
• time to peak of HDR

75
Long term plans

• Use MRI to monitor structural (growth, cortical
  folding, laying down of white matter tracts) and
  functional brain development in normal fetuses
• Determine changes in these parameters in
  compromised fetuses.
• Determine whether abnormalities in fetal brain
  activity or structural development relate to
  later neurological deficit