- Email: [email protected]
Respiratory triggered imaging with an optical displacement sensor
Magnetic
Kesononce
Imaging,
Vol.
II,
pp.
1027-1032,
1993
0730-725X/93 $6.00 + .oO Copyright 0 1993 Pergamon Press Ltd.
Printed in the USA. All rights reserved.
l Technical Note
RESPIRQTORY TRIGGERED IMAGING WITH AN OPTICAL DISPLACEMENT SENSOR STEPHEN J. WILSON, IAN M. BRERETON, PAUL HOCKINGS, WOLFGANG ROFFMANN, AND DAVID M. DODDRELL Centre for Magnetic Resonance, University of Queensland, St. Lucia 4072, Australia Motion of abdominal organs with respiration is a major problem in NMR spectroscopy and imaging thereof. Triggering each phase-encoding step with respiration or gating a number of phase-encoding steps is one approach to the problem. The design of a sensor for small animal experiments has not been as simple. An optical device, implemented with polymer optical fibres is described, along with associated hardware and electronics which can act as a trigger for small animal NMR experiments. A brief description of a similar device for human application is also given. 2DFT spin-echo and B. susceptibility images, both triggered and untriggered, are presented to validate the technique. Kepords:
MRI; Motion artefact suppression; Respiratory sensor; Mouse anatomy.
for routine imaging. Respiratory ordering of phase encoding (ROPE),4 to reduce ghosting, is a well-used technique in human imaging. Periodic motion in the phase-encoding direction is made to appear as a monotonic function in k-space leading to the absence of ghosts. The image may also be acquired over a small period of time relative to the motion using techniques such as FLASH,’ and so on. However, there is a concomitant degradation of the signal-to-noise ratio (SNR). Alternatively excitation and acquisition can be synchronized to a period of minimal motion during the respiration cycle of a subject breathing normally. While movement may still occur during acquisition, variation between phase-encoding steps is minimized leading to reduced image ghosting. In general, respiratory gated imaging requires the radiofrequency (RF) excitation and acquisition to be triggered by a signal from a sensor device, sensitive to the respiration cycle. A triggered pulse sequence increments one phase encoding step per trigger pulse (which can preclude T1-weighted images depending on the respiratory rate of the subject). A gated sequence allows one or more phase-encoding steps during the gating pulse, which is active during a relatively stationary phase of the respiratory cycle. In this manner, flexibility as to the setting of TR is afforded. Other than directly from a ventilator,3 this signal has been derived from vari-
INTRODUCTION
Both magnetic resonance (MR) imaging and spectroscopic investigation of the abdomen suffer quality degradation due to respiratory motion. Periodic movement during MR imaging data collection produces ghosting in the phase-encoding direction, which leads to loss of image definition and accuracy of quantitative measurement. Localized MR spectroscopy is equally susceptible to motion effects that interfere with local shimming optimization and induce error in localization, particularly in those methods relying on signal cancellation for voxel selection. A number of techniques have been developed to minimize motional artifacts in MR imaging,’ including respiration control (either voluntary or mechanical), reordering of phase-encoding steps, controlled timing (gating) of image acquisition, rapid image acquisition, and image post-processing.2 Respiration control involves data acquisition while the abdomen is held stationary following expiration, and may be achieved by breath holding or mechanical ventilation. Breath holding is clearly not applicable to animal studies or human neonatal or pediatric imaging. Data acquisition synchronized to the operation of a ventilator has been used to suppress breathing artifacts in images of the rat abdomen,3 and may be applied to human subjects under anaesthesia, but it is not ideal RECEIVED
10/2/92;
ACCEPTED
3/16/93. 1027
Magnetic Resonance Imaging 0 Volume 1I, Number 7, 1993
1028
ous sources. Physical monitoring of breathing motion by measuring pressure changes induced in liquid or airfilled balloons or jackets placed on the subject has been employed,6 but by restricting chest or abdominal excursions, these techniques compromise ventilation. Movement in the magnetic field by coils or wires positioned on the chest also provides a triggering mechanism.’ Thermistors may be used to monitor changes in breath temperature on exhalation, but are difficult to set up and run over the required time periods, particularly in view of ambient temperature changes. Being fabricated of magnetic or paramagnetic materials, the two latter techniques also compromise field homogeneity when placed in proximity of the field of interest. Here we describe the use of an optical fibre sensor for gated acquisition of abdominal images of the mouse. The device is convenient to operate, suitable for small animal and human application (Fig. l), imposes no physical load upon the subject, can be adjusted while outside the magnet, and provides no electrical interference. MATERIALS
AND METHODS
Optical motion detectors take many forms and are particularly suitable where a degree of isolation, either physical or electrical, from the measured subject is required. The workload required to operate an optical sensor is minimal, allowing greater accuracy and sensitivity. A 660 nm LED (red) served as a visible light source and, with an integral lens, was coupled to optical fibre.
The sensor consisted of a length (3-6 cm depending on the subject) of flexible vinyl tube (see Appendix for supplier) with an internally reflective surface attached to the abdominal wall. Deformation of the tube during respiratory excursions reduces internal reflection of light launched into one end, modulating the intensity emitted from the other end. A phototransistor detected this signal and after amplification, filtering and comparison with a threshold, a trigger signal compatible with the spectrometer was provided, as depicted schematically in Fig. 2A. A presetable trigger delay allowed delays from 0 to 700 msec (Fig. 2B). The source, sensor and detector were coupled with 1 mm single core, step index polymer optical cable. Signal attenuation of 150 dB/km at 660 nm was insignificant over 3-4 metre cable runs. Four-metre lengths of optical fibre allowed the electronics to be housed outside of the magnet room minimising stray RF interference. Experimental protocol for the use of the optical motion detector was to anaesthetise the animal (1.8% isoflurane in oxygen, 300 ml/min), shave a small area of fur on the median line is order to adhere the sensor tube to the animal at one point. Cyanoacrylate adhesive was used which provided a biocompatible bond to the abdominal wall.* The sensor tube was shed by the animal after approximately 8 hr. The source and detector fibres, fixed to an animal bed, were then pushfitted to the sensor tube and mounted at the head and tail of the animal, respectively, as shown in Fig. 3. MR images were acquired on a modified Bruker MSL200 spectrometer under Tomikon software con-
E\dhesive
d Source
and
s!lrface
detector
fibrcs
Fig. 1. The form of the sensor for human use (shown) differs from that described for small animal use. Two sensor tubes are positioned at the pivot point of an optical fibre loop which is supported by two rigid members. An angular displacement sensor is formed and when applied to the lower costal margin the rise and fall of the abdomen modulates the intensity of transmitted light. Conversion of this optical signal to a voltage and triggering is then performed.
:
Dl
?
R
LIl358 LH311 WE555 4R2B
IC5 WE555
8 nsec
pulse
Fig. 2A. A schematic diagram of the detecting electronics used to trigger the spectrometer 5%). Fibre optic parts suppliers are listed in the Appendix.
IClr2 IC3 IC4,5 IC6
660 nr. 8OUrCe
R0 330
4
t
+ f
ICZ/A
_
Optical
at a fixed point in the respiratory
Threshold
trigger
cycle. All resistors
NMR
A
Thres.
Delay
C6
are metal film (i W,
B2
Bl
NE555
IC4
Magnetic Resonance Imaging 0 Volume 1I, Number 7, 1993
1030
..._.DE!!.?!_ . ..d .I T
i
Fig. 2B. A typical respirogram of the animal indicating the triggering point of the comparator(T) and the use of a hardware delay to place the trigger pulse in a stationary region of the respiratory cycle (T’).
trol, interfaced to a horizontal 40 cm bore Spectrospin magnet system operating at 4.5 T. Homebuilt birdcage resonator and gradient set of internal diameters 70 and 100 mm, respectively, were used. Adult Wistar rats and Quackenbush mice were used to access the operation of the respiratory trigger device. Both standard 2DFT spin-echo images and B0 field maps9 acquired with a sequence incorporating a frequency selective l-1 pulse in the absence of a gradient followed by a slice-selective 180” pulse, were acquired
Fig. 3. The animal is shown ready for insertion into the resonator. Source and detector fibres are seen entering from the left with the sensor tube applied to the upper abdomen.
(4
(W
Fig. 4. A transverse section through the liver of an adult mouse using MSME.TR = 900 msec, TE = 18 msec, slice thickness 2 mm. The untriggered image (A) shows characteristic ghosting and blurring due to motion of abdominal organs with respiration. The triggered image (B) is devoid of ghosting and shows clearly the vascular structure of the liver.
Respiratory triggered imaging 0 S.J.
WILSON
1031
ET AL.
@I
(4 Fig. 5. The magnetic susceptibility
map of the mouse untriggered (A) shows poor definition of structure and isofrequency lines, whereas the triggered counterpart (B) shows abdominal organs, gut, and kidneys with obvious isofrequency lines.
with and without respiratory triggering (TE = 18 msec in both cases). For the non-triggered images, TR was set to match, as closely as possible, the effective recycle delay induced by the respiration period in the triggered images. RESULTS
AND DISCUSSION
The procedure outlined above was used to obtain respiratory gated MR images from adult rats and mice. A comparison of normal and gated transverse spinecho images of the abdomen of a mouse is shown in Fig. 4. The untriggered image displays the characteristic ghosting associated with chest movement, while image ghosting is reduced in the respiratory triggered image. Cardiac motion is responsible for the residual artefacts which could be reduced via additional cardiac gating. The improvement in image definition with respiratory triggering is also demonstrated in Fig. 5. Overcoming the problem of motional triggering in the NMR environment is one of the major obstacles to spectroscopy and imaging studies of abdominal and thoracic organs. The optical technique described here presents a simple, robust solution to the problem of respiratory triggering in small animal NMR experiments. APPENDIX:
PARTS
A. Vinyl tubing, No. 2, Portex Hythe, Kent CT216JL.
LIST Ltd,
High
Street,
B. Optical Fibre, Duplex Polymer Cable, Cat. 368053, RS Components. C. Emitter and Detector Housing, Cat. 456-605, RS Components. D. Optical fibre plug, Cat. 456-598, RS Components. E. Emitter, GaAsP LED SE4355, Honeywell Optoelectronics, 830 East Arapaho Road, Richardson, Texas 7508 1. F. Detector, Photodarlington MEL12, Microelectronics, 38 Hung To Road, Kwun Tong, Kowloon, Hong Kong. REFERENCES 1. Wood, M.L.; Henkelman, R.M. Suppression of respiratory motion artefacts in magnetic resonance imaging. Med. Phys. 13:794-805; 1986. 2. Hedley, M.; Yan, H. Motion artifact suppression: A review of post-processing techniques. Magn. Reson. Imaging 10:627-635; 1992. 3. Hedlund, L.W.; Johnson, G.A.; Mills, G.I. Magnetic resonance microscopy of the rat thorax and abdomen. Invest. Radiof. 21:843-846; 1986. 4. Bailes, D.R.; Gilderdale, G.M.; Bydder, G.M.; Collins, A.G.; Firmin, D.N. Respiratory ordered phase encoding (ROPE): A method for reducing respiratory motion artefacts in MR imaging. J. Comput. Assist. Tomogr. 9(4): 835-838; 1985. 5. Haase, A.; Frahm, J.; Matthaei, D.; Manicke, W. Merboldt, K.D. FLASH imaging rapid NMR imaging using low flip-angle pulses. J. Magn. Reson. 67:258-266; 1986.
1032
Magnetic Resonance Imaging 0 Volume 11, Number 7, 1993
6. Brauer, M.; Towner, R.A.; Renaud, I.; Janzen, E.G.; Foxall, D.I. In vivo proton nuclear magnetic resonance imaging and spectroscopy studies of halocarbon-induced liver damage. Magn. Reson. Med. 9:229; 1989. 7. Van Bruggen, N.; Syha, J.; Busza, A.L.; King, M.D.; Stamp, G.W.H.; Williams, S.R.; Gradian, D.G. Identification of tumour haemorrhage in an animal model using
spin echoes and gradient echoes. Magn. Reson. Med. 15: 121-127; 1990. 8. Editorial: Cyanoacrylate tissue adhesives. JAMA 201: 113; 1967.
9. Jung, W.I.; Kiiper, K.; Lutz, 0.; Miiller, K.; Pfeffer, M. Magnetic field-mapping by multi-slice MAGNEX. App. Magn. Reson. 1:497-507; 1990.
Kesononce
Imaging,
Vol.
II,
pp.
1027-1032,
1993
0730-725X/93 $6.00 + .oO Copyright 0 1993 Pergamon Press Ltd.
Printed in the USA. All rights reserved.
l Technical Note
RESPIRQTORY TRIGGERED IMAGING WITH AN OPTICAL DISPLACEMENT SENSOR STEPHEN J. WILSON, IAN M. BRERETON, PAUL HOCKINGS, WOLFGANG ROFFMANN, AND DAVID M. DODDRELL Centre for Magnetic Resonance, University of Queensland, St. Lucia 4072, Australia Motion of abdominal organs with respiration is a major problem in NMR spectroscopy and imaging thereof. Triggering each phase-encoding step with respiration or gating a number of phase-encoding steps is one approach to the problem. The design of a sensor for small animal experiments has not been as simple. An optical device, implemented with polymer optical fibres is described, along with associated hardware and electronics which can act as a trigger for small animal NMR experiments. A brief description of a similar device for human application is also given. 2DFT spin-echo and B. susceptibility images, both triggered and untriggered, are presented to validate the technique. Kepords:
MRI; Motion artefact suppression; Respiratory sensor; Mouse anatomy.
for routine imaging. Respiratory ordering of phase encoding (ROPE),4 to reduce ghosting, is a well-used technique in human imaging. Periodic motion in the phase-encoding direction is made to appear as a monotonic function in k-space leading to the absence of ghosts. The image may also be acquired over a small period of time relative to the motion using techniques such as FLASH,’ and so on. However, there is a concomitant degradation of the signal-to-noise ratio (SNR). Alternatively excitation and acquisition can be synchronized to a period of minimal motion during the respiration cycle of a subject breathing normally. While movement may still occur during acquisition, variation between phase-encoding steps is minimized leading to reduced image ghosting. In general, respiratory gated imaging requires the radiofrequency (RF) excitation and acquisition to be triggered by a signal from a sensor device, sensitive to the respiration cycle. A triggered pulse sequence increments one phase encoding step per trigger pulse (which can preclude T1-weighted images depending on the respiratory rate of the subject). A gated sequence allows one or more phase-encoding steps during the gating pulse, which is active during a relatively stationary phase of the respiratory cycle. In this manner, flexibility as to the setting of TR is afforded. Other than directly from a ventilator,3 this signal has been derived from vari-
INTRODUCTION
Both magnetic resonance (MR) imaging and spectroscopic investigation of the abdomen suffer quality degradation due to respiratory motion. Periodic movement during MR imaging data collection produces ghosting in the phase-encoding direction, which leads to loss of image definition and accuracy of quantitative measurement. Localized MR spectroscopy is equally susceptible to motion effects that interfere with local shimming optimization and induce error in localization, particularly in those methods relying on signal cancellation for voxel selection. A number of techniques have been developed to minimize motional artifacts in MR imaging,’ including respiration control (either voluntary or mechanical), reordering of phase-encoding steps, controlled timing (gating) of image acquisition, rapid image acquisition, and image post-processing.2 Respiration control involves data acquisition while the abdomen is held stationary following expiration, and may be achieved by breath holding or mechanical ventilation. Breath holding is clearly not applicable to animal studies or human neonatal or pediatric imaging. Data acquisition synchronized to the operation of a ventilator has been used to suppress breathing artifacts in images of the rat abdomen,3 and may be applied to human subjects under anaesthesia, but it is not ideal RECEIVED
10/2/92;
ACCEPTED
3/16/93. 1027
Magnetic Resonance Imaging 0 Volume 1I, Number 7, 1993
1028
ous sources. Physical monitoring of breathing motion by measuring pressure changes induced in liquid or airfilled balloons or jackets placed on the subject has been employed,6 but by restricting chest or abdominal excursions, these techniques compromise ventilation. Movement in the magnetic field by coils or wires positioned on the chest also provides a triggering mechanism.’ Thermistors may be used to monitor changes in breath temperature on exhalation, but are difficult to set up and run over the required time periods, particularly in view of ambient temperature changes. Being fabricated of magnetic or paramagnetic materials, the two latter techniques also compromise field homogeneity when placed in proximity of the field of interest. Here we describe the use of an optical fibre sensor for gated acquisition of abdominal images of the mouse. The device is convenient to operate, suitable for small animal and human application (Fig. l), imposes no physical load upon the subject, can be adjusted while outside the magnet, and provides no electrical interference. MATERIALS
AND METHODS
Optical motion detectors take many forms and are particularly suitable where a degree of isolation, either physical or electrical, from the measured subject is required. The workload required to operate an optical sensor is minimal, allowing greater accuracy and sensitivity. A 660 nm LED (red) served as a visible light source and, with an integral lens, was coupled to optical fibre.
The sensor consisted of a length (3-6 cm depending on the subject) of flexible vinyl tube (see Appendix for supplier) with an internally reflective surface attached to the abdominal wall. Deformation of the tube during respiratory excursions reduces internal reflection of light launched into one end, modulating the intensity emitted from the other end. A phototransistor detected this signal and after amplification, filtering and comparison with a threshold, a trigger signal compatible with the spectrometer was provided, as depicted schematically in Fig. 2A. A presetable trigger delay allowed delays from 0 to 700 msec (Fig. 2B). The source, sensor and detector were coupled with 1 mm single core, step index polymer optical cable. Signal attenuation of 150 dB/km at 660 nm was insignificant over 3-4 metre cable runs. Four-metre lengths of optical fibre allowed the electronics to be housed outside of the magnet room minimising stray RF interference. Experimental protocol for the use of the optical motion detector was to anaesthetise the animal (1.8% isoflurane in oxygen, 300 ml/min), shave a small area of fur on the median line is order to adhere the sensor tube to the animal at one point. Cyanoacrylate adhesive was used which provided a biocompatible bond to the abdominal wall.* The sensor tube was shed by the animal after approximately 8 hr. The source and detector fibres, fixed to an animal bed, were then pushfitted to the sensor tube and mounted at the head and tail of the animal, respectively, as shown in Fig. 3. MR images were acquired on a modified Bruker MSL200 spectrometer under Tomikon software con-
E\dhesive
d Source
and
s!lrface
detector
fibrcs
Fig. 1. The form of the sensor for human use (shown) differs from that described for small animal use. Two sensor tubes are positioned at the pivot point of an optical fibre loop which is supported by two rigid members. An angular displacement sensor is formed and when applied to the lower costal margin the rise and fall of the abdomen modulates the intensity of transmitted light. Conversion of this optical signal to a voltage and triggering is then performed.
:
Dl
?
R
LIl358 LH311 WE555 4R2B
IC5 WE555
8 nsec
pulse
Fig. 2A. A schematic diagram of the detecting electronics used to trigger the spectrometer 5%). Fibre optic parts suppliers are listed in the Appendix.
IClr2 IC3 IC4,5 IC6
660 nr. 8OUrCe
R0 330
4
t
+ f
ICZ/A
_
Optical
at a fixed point in the respiratory
Threshold
trigger
cycle. All resistors
NMR
A
Thres.
Delay
C6
are metal film (i W,
B2
Bl
NE555
IC4
Magnetic Resonance Imaging 0 Volume 1I, Number 7, 1993
1030
..._.DE!!.?!_ . ..d .I T
i
Fig. 2B. A typical respirogram of the animal indicating the triggering point of the comparator(T) and the use of a hardware delay to place the trigger pulse in a stationary region of the respiratory cycle (T’).
trol, interfaced to a horizontal 40 cm bore Spectrospin magnet system operating at 4.5 T. Homebuilt birdcage resonator and gradient set of internal diameters 70 and 100 mm, respectively, were used. Adult Wistar rats and Quackenbush mice were used to access the operation of the respiratory trigger device. Both standard 2DFT spin-echo images and B0 field maps9 acquired with a sequence incorporating a frequency selective l-1 pulse in the absence of a gradient followed by a slice-selective 180” pulse, were acquired
Fig. 3. The animal is shown ready for insertion into the resonator. Source and detector fibres are seen entering from the left with the sensor tube applied to the upper abdomen.
(4
(W
Fig. 4. A transverse section through the liver of an adult mouse using MSME.TR = 900 msec, TE = 18 msec, slice thickness 2 mm. The untriggered image (A) shows characteristic ghosting and blurring due to motion of abdominal organs with respiration. The triggered image (B) is devoid of ghosting and shows clearly the vascular structure of the liver.
Respiratory triggered imaging 0 S.J.
WILSON
1031
ET AL.
@I
(4 Fig. 5. The magnetic susceptibility
map of the mouse untriggered (A) shows poor definition of structure and isofrequency lines, whereas the triggered counterpart (B) shows abdominal organs, gut, and kidneys with obvious isofrequency lines.
with and without respiratory triggering (TE = 18 msec in both cases). For the non-triggered images, TR was set to match, as closely as possible, the effective recycle delay induced by the respiration period in the triggered images. RESULTS
AND DISCUSSION
The procedure outlined above was used to obtain respiratory gated MR images from adult rats and mice. A comparison of normal and gated transverse spinecho images of the abdomen of a mouse is shown in Fig. 4. The untriggered image displays the characteristic ghosting associated with chest movement, while image ghosting is reduced in the respiratory triggered image. Cardiac motion is responsible for the residual artefacts which could be reduced via additional cardiac gating. The improvement in image definition with respiratory triggering is also demonstrated in Fig. 5. Overcoming the problem of motional triggering in the NMR environment is one of the major obstacles to spectroscopy and imaging studies of abdominal and thoracic organs. The optical technique described here presents a simple, robust solution to the problem of respiratory triggering in small animal NMR experiments. APPENDIX:
PARTS
A. Vinyl tubing, No. 2, Portex Hythe, Kent CT216JL.
LIST Ltd,
High
Street,
B. Optical Fibre, Duplex Polymer Cable, Cat. 368053, RS Components. C. Emitter and Detector Housing, Cat. 456-605, RS Components. D. Optical fibre plug, Cat. 456-598, RS Components. E. Emitter, GaAsP LED SE4355, Honeywell Optoelectronics, 830 East Arapaho Road, Richardson, Texas 7508 1. F. Detector, Photodarlington MEL12, Microelectronics, 38 Hung To Road, Kwun Tong, Kowloon, Hong Kong. REFERENCES 1. Wood, M.L.; Henkelman, R.M. Suppression of respiratory motion artefacts in magnetic resonance imaging. Med. Phys. 13:794-805; 1986. 2. Hedley, M.; Yan, H. Motion artifact suppression: A review of post-processing techniques. Magn. Reson. Imaging 10:627-635; 1992. 3. Hedlund, L.W.; Johnson, G.A.; Mills, G.I. Magnetic resonance microscopy of the rat thorax and abdomen. Invest. Radiof. 21:843-846; 1986. 4. Bailes, D.R.; Gilderdale, G.M.; Bydder, G.M.; Collins, A.G.; Firmin, D.N. Respiratory ordered phase encoding (ROPE): A method for reducing respiratory motion artefacts in MR imaging. J. Comput. Assist. Tomogr. 9(4): 835-838; 1985. 5. Haase, A.; Frahm, J.; Matthaei, D.; Manicke, W. Merboldt, K.D. FLASH imaging rapid NMR imaging using low flip-angle pulses. J. Magn. Reson. 67:258-266; 1986.
1032
Magnetic Resonance Imaging 0 Volume 11, Number 7, 1993
6. Brauer, M.; Towner, R.A.; Renaud, I.; Janzen, E.G.; Foxall, D.I. In vivo proton nuclear magnetic resonance imaging and spectroscopy studies of halocarbon-induced liver damage. Magn. Reson. Med. 9:229; 1989. 7. Van Bruggen, N.; Syha, J.; Busza, A.L.; King, M.D.; Stamp, G.W.H.; Williams, S.R.; Gradian, D.G. Identification of tumour haemorrhage in an animal model using
spin echoes and gradient echoes. Magn. Reson. Med. 15: 121-127; 1990. 8. Editorial: Cyanoacrylate tissue adhesives. JAMA 201: 113; 1967.
9. Jung, W.I.; Kiiper, K.; Lutz, 0.; Miiller, K.; Pfeffer, M. Magnetic field-mapping by multi-slice MAGNEX. App. Magn. Reson. 1:497-507; 1990.