HAL Id: inserm-00921893
https://www.hal.inserm.fr/inserm-00921893
Submitted on 7 Jul 2014
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
An optical ber-based gating device for prospective
mouse cardiac MRI.
Raphaël Sablong, Adrian Rengle, Anoop Ramgolam, Hervé Saint-Jalmes,
Olivier Beuf
To cite this version:
Raphaël Sablong, Adrian Rengle, Anoop Ramgolam, Hervé Saint-Jalmes, Olivier Beuf. An
optical ber-based gating device for prospective mouse cardiac MRI.. IEEE Transactions on
Biomedical Engineering, Institute of Electrical and Electronics Engineers, 2014, 61 (1), pp.162-70.
�10.1109/TBME.2013.2278712�. �inserm-00921893�
An optical fiber-based gating device for
prospective mouse cardiac MRI
R. Sablong, A. Rengle, A. Ramgolam, H. Saint-Jalmes and O. Beuf
Abstract—Prospective synchronization of MRI acquisitions on living organisms involves the monitoring of
respiratory and heart motions. The electrocardiogram (ECG) signal is conventionally used to measure the cardiac
cycle. However, in some circumstances, obtaining an uncorrupted ECG signal recorded on small animals with RF
pulses and gradient switching is challenging. To monitor respiratory motion, an air cushion associated with a pressure
sensor is commonly used but the system suffers from bulkiness. For many applications, the physiological gating
information can also be derived from an MR navigated-signal. However, a compact device that can simultaneously
provide respiratory and cardiac information, for both prospective gating and physiological monitoring, is desirable.
This is particularly valid since small volume coils or dedicated cardiac RF coil arrays placed directly against the chest
wall are required to maximize measurement sensitivity. An optic-based device designed to synchronize MRI
acquisitions on small animal’s respiratory and heart motion was developed using a transmit-receive pair of optical
fibers. The suitability of the developed device was assessed on mice (n=10) and was based on two sets of experiments
with dual cardiac and respiratory synchronization. Images acquired with prospective triggering using the optical-
based signal, ECG and the pressure sensor during the same experiment were compared between themselves in the
first set. The second set compared prospective technique using optical-based device and ECG to a retrospective
technique. The optical signal which was correlated to both respiratory and heart motion was totally unaffected by
radiofrequency pulses or currents induced by the magnetic field gradients used for imaging. Mice heart MR images
depict low visible motion artifacts with all sensors or techniques used. No significant SNR differences were found
between each series of image. Full fiber optic based signal derived from heart and respiratory motion was suitable for
prospective triggering of heart MR imaging. The fiber optic device performed similarly to the ECG and air pressure
sensors, while providing an advantage for imaging with dedicated cardiac array coils by reducing bulk. It can be an
attractive alternative for small animal MRI in difficult environments such as limited space and strong gradient
switching.
Index Terms—Cardio-respiratory triggering, optical sensor fibers, prospective, retrospective triggering, mouse
heart MRI.
I. INTRODUCTION
There has been a rising interest for small animal Magnetic Resonance Imaging and Spectrometry (MRI-MRS)
since many years. The growing number of animal models and recent advances in dedicated NMR instruments are
contributing to the current abundance of studies on the vascular, cardiac and abdominal systems as well as on the
brain, typically in mice and rats. In most cases, suitable images and spectra are achieved through the
implementation of triggering methods. Indeed, movements of the rib cage (ventilation) and the heart (heartbeat)
generate artifacts that can significantly impair the quality of acquired images and spectra, especially when high
spatial resolution is required [1]. These movements are not only due to the heart and lungs, but also have a
systemic character. Artifacts appear mainly as localized blurring or ghosting propagating along the phase
encoding direction in imaging and as line broadening in spectroscopy. The extension and intensity of these
movement artifacts depend on various parameters such as physiological conditions (anesthesia, species...),
mechanical properties (studied organ, volume available...) and acquisition constraints (spatial or temporal
resolution, sequence used…). Several strategies were implemented to mitigate the effect of these movements.
Different methods using prospective cardio-respiratory synchronizations are widely used for acquisitions of
this kind. The NMR signal is acquired only at appropriate timings concurring to a real-time signal obtained from
the measurement of the animal’s movements due to breathing and heartbeat cycle. The former is composed of a
fast inspiration phase and a slow end expiration phase. One of the ways of differentiating these two phases
during which the acquisition can take place, is to use a pneumatic system where a balloon is placed on the
animal’s abdomen and allows respiratory movement monitoring through pressure fluctuations. Another common
technique consists of using an Electrocardiogram (ECG). The electro-physiological signal is an accurate
signature of the heart beat with successive waves defining the different heart phases (diastolic, systolic); this
either allows an acquisition to be performed at a specific phase of the heart beat or to acquire several frames of
the heart cycle followed by CINE movies reconstruction. However, Radio frequency (RF) field and fast gradient
switching can induce currents along the ECG leads that can potentially disturb the ECG signal and additionally
induce localized heating of the region sporting the electrodes. Non-metallic materials like carbon wires have
been proposed almost 25 years ago [2] and is still an active research area [3]. While carbon leads are reducing
heating, the use of carbon wires alone is not sufficient to solve induced ECG artifacts in worst cases. Such
approach has to be combined with an electro-optical conversion module to be really efficient. A second electro-
optical conversion module placed at a convenient distance from the magnet converts the optical signal into an
electrical signal that can be processed by the triggering device [4]. An ECG-based device often spares the use of
an additional breathing sensor by using the modulation of the ECG by a low frequency wave associated with
ventilation. Specific signal processing allows discriminating the two physiological signal sources and thus to
perform a dual cardio-respiratory synchronization.
Several studies [5, 6] reported challenging cardiac synchronization using ECG especially for cardiac
microscopy when short repetition time and high gradient-slew rate are used. Quite a few studies have sorted
several alternatives and tested them successfully. Respiratory-gated and cardiac-triggered spin-echo images of
the rat abdomen and mouse heart were performed with an inductive pickup coil placed on the animal's chest
instead of using standard ECG leads [7] but the method is not optimal for use with RF surface coils or arrays for
cardiac imaging. Other straightforward optical detection systems, based on the interruption of an infrared optical
beam, have previously been proposed for respiratory gating only [8]. Detection without contact, based on the
infrared reflectometry principle has previously been characterized by Lemieux [9]. Different authors have used a
dedicated bed to facilitate the non-invasive placement of an optical probe coupled to optical fibers for mice liver
examination at 4.7T [10] and 7T [11]. The ease of use and low cost of the fiber-optic detector was further
demonstrated on mouse liver by enabling respiratory-synchronized 1H MR Spectroscopy acquisition [12].
Studies based on optical detection to measure the cardiac motion are scarce. Only a handful of studies
pertaining to few invasive or minimally invasive methods unaffected by RF fields and gradient switching can be
found in the literature. An invasive miniature optical probe inserted into an artery showed an optical signal
resulting from a fluctuating light reflection caused by the pulsatile blood flow [13]. A second study described an
optical stethoscope where an optical fiber was carefully introduced into the animal’s esophagus and the fiber tip
then placed close to the mouse’s heart. A signal correlated to the cardiac cycle was then measured via this
setup [14]. However, positioning the probe at the exact location requires expertise and hence limits the
widespread diffusion of such minimally invasive technique. More recently, the reliability of a cardio-acoustic
triggering device for human cardiac CINE imaging at 7T [15] as well as the feasibility to use
MagnetoHydroDynamic (MHD) effects for synchronization of MR acquisitions with the cardiac cycle [16] were
demonstrated.
On another hand, retrospective techniques based on navigated-signal do not require any sensors. However
they are not compatible with all acquisitions. Rectilinear self-navigated motion detection techniques were first
proposed for human cardiac imaging [17, 18] and extended to abdominal imaging [19]. This technique
eliminates the need for an ECG or respiratory signal by recording a motion synchronization signal directly from
non-triggered navigator data. A navigated signal, acquired without phase or readout gradients, is used to achieve
retrospective gating ahead of image reconstruction.
Abiding to the context, this study aims at comparing the prospective triggering techniques with a retrospective
technique. Three types of sensors capable of monitoring the heart beat and the breathing cycle were used: (a)
reflected light modulation using an original device with optical fibers, (b) ECG, (c) pressure sensor via air
cushion. Images were obtained from a short axis-orientation mouse cardiac CINE FLASH sequence and
compared between themselves. The goal of this study was to demonstrate that optical fiber is a suitable
alternative for dual cardio-respiratory gating without the limitations of ECG or respiratory pillow.
II. EXPERIMENTAL
A. Optical gating device
The optical probe consists of a bundle of two silica multimode optical fiber of 200 μm core diameter (one for
light transmission and one for light detection). Distal parts were stripped (2 cm long), cleaved and polished to
maximize light detection and the two fibers were afterwards glued together. Each fiber was optically insulated
with thin heat shrink cover to minimize the ambient light noise. Once assembled, the optical probe is a thin and
soft cylindrical line of less than 2mm diameter with a junction at about 5 cm apart from the tip enlarging
punctually the diameter to 3 mm. The proximal part of the transmit fiber was connected to a 820 nm wavelength
100 μW transmit power HFBR-1405 light emitting diode (LED) whereas the detection fiber was connected to a
HFBR-2405 light-voltage amplified photodiode receiver (Agilent Technologies, CA, USA). The transmit fiber
was used to illuminate the moving surface and the receiver fiber captured a fraction of the backscattered light.
This quantity of light is modulated by the cardio-respiratory movements of the animal’s chest. After conversion
of the modulated light into an electrical signal, the latter was passed through a custom-built signal-processing
circuit for further amplification and filtering. This circuit consisted of two active wide band pass filter placed
before and after an amplifier with adjustable gain. Each active wide band pass circuit was composed of two
active second order high pass filters and two active second order low pass filters. The corner frequencies were
0.2 Hz and 30 Hz respectively. The repetition rate of the mice’s heart beat is typically about 5 to 10 Hz while
that of the ventilation movement varies between 0.5 to 1 or 2 Hz. These physiological frequencies are included
in the filter passband. Any continuous component of the signal is filtered out and the upper cutt-off frequency,
together with a fourth order filter profile, enables to greatly reject 50 Hz residues due to any parasitic galvanic
coupling to power supply. The filter design was based on a Sallen-Key topology. The amplifier stage with
adjustable gain was conceived using an operational amplifier in an inverting configuration with a
100/1,000/10,000 scale for an adjustable gain. A fixed gain of 10,000 was selected. With a narrow bandwidth
(30 Hz) and the use of low noise operational amplifiers for active filtering and amplification, high signal-to-noise
detection can be performed. The processed output signal was then connected to a commercial Trigger Unit HR
V2.0 (Rapid Biomedical, Würzburg, Germany) for gating purpose (Fig. 1).
B. Optical gating device characterization
The characteristics of the optical-based device were measured using a calibrated light source. A light intensity
modulated laser (670 nm, 1 mW average power, Laser Components GMBH, Germany) was used to illuminate
the fiber tip all while the beam being attenuated by a white screen to avoid detector saturation. The device
transfer function was assessed frequency-wise. The laser intensity modulation frequency was controlled by
means of a standard 33220A low frequency generator in sweep mode (Agilent technologies, CA, USA). The
input and output signals were both recorded using a data acquisition board with Labview® Software Interface
(National Instruments, USA). Thus to check whether the measured Bode diagram fits the expected
specifications, the harmonic response of the device is accurately determined in terms of amplitude and phase.
Besides, the transient response to a typical input signal was also investigated: the impact of the frequential
characteristics of the filter on the waveform of the signal corresponding to mouse monitoring is better illustrated
in the time domain. Indeed the waveform of the measured signal is determined by both the optical and
mechanical processes (which characterize light propagation within a complex medium, in our case the tissue)
and the bandpass of the filter. The latter being quite narrow (30 Hz), the output signal is inevitably distorted and
thus denies access to the “real” input signal. Therefore an approximated input signal in the form of an ECG-like
waveform mirroring the heart activity was used to illustrate the distortions induced by the filter. This signal
(since the input signal is unknown) was generated using the 33220A generator set in arbitrary waveform mode to
drive the modulated laser intensity. An ECG-like signal was generated from a specific 512 data points file (.csv
format) coded in Matlab (http://www.mathworks.com/matlabcentral/fileexchange/10858-ecg-simulation-using-
matlab) using the template described in reference [20]. The main waveform features were the P,Q,R,S,T waves
from which the amplitude, shape and phase were set to be comparable to the typical physiological parameters of
mouse heart beats. The repetition rate of this temporal pattern was set to be 6.6 Hz.
The optical device response was compared to the driving signal. Finally an optical power meter enabled to
measure the minimum and maximum intensity in milliwatts of the modulated light source.
C. Experimental in vivo setup
Triggering efficiency was assessed through image quality acquired on ten OF1 mice (6 weeks old with 26±2 g
average weight). The first series was dedicated to prospective triggering with successive acquisitions performed
on each mouse using the optical, ECG and pressure signals respectively. The optical fiber pair was first fixed on
the thorax using soft medical adhesive tape. For ECG, the front paws were wrapped in copper foil and the
peripheral ECG signal was derived via silver wire. To minimize both corrupted signal and heating due to
gradient switching [21], the loop formed by the paws and the cable ends was kept as small as possible by
twisting both wires. The screened 2-wire ECG cable was then guided straight in the z-direction through the
magnet close to the symmetry axis. Special care was taken in these experimental conditions not to be in a worst
case scenario and avoid ECG signal failure with the commercial triggering unit available at the laboratory.
Finally, the air cushion was placed above and taped on the fiber tips (Fig. 2). The air cushion was a 20 mm
diameter and 3 mm thick soft disk connected by a 2 mm outer diameter catheter to a deported pressure sensor.
Approximately the same delays, low and high cutoff frequencies for filtering were applied on the ECG trigger
unit. Typical signals from these sensors have been recorded to show that the different devices provide
comparable real-time information about cardiac and breathe movements of a given animal. Two days later, a
second series of acquisitions on the same mice was performed with prospective triggering using optical and ECG
sensors and a retrospective triggering sequence. The experimental protocol was approved by the Animal Ethics
Committee of our institution and ethical guidelines for experimental investigations with animals were strictly
followed. A dedicated anesthesia system (TEM, Lormont, France) using isoflurane gas was used to perform the
anesthesia. Induction was realized with 4% gas mixed with air administrated at 1°L/min flow. The animals were
placed in a supine position on a dedicated plastic bed with circulating warm water for body temperature
regulation while anesthesia was maintained during MRI examination with 2% isoflurane mixed with air at 0.6 to
1 L/min flow. A capillary filled with a 1.25 g/L NiSO4 solution and placed below the mouse was used as an
external image reference.
D. Imaging protocol
Experiments were performed on a 4.7T Biospec system (Bruker, Ettlingen, Germany) with a quadrature 32 mm
inner diameter birdcage coil (Rapid Biomedical, Würzburg, Germany). Short axis-orientation images of the heart
were obtained using a CINE FLASH sequence with the following parameters: 30 x 30 mm
2
field-of-view (FOV),
256x192 matrix, 4 averages; phase anti-aliasing=2; TR/TE=9/2.9 ms; 25° flip angle; 1 mm slice thickness; 50
kHz receive bandwidth. With an average heart rate of 400 bpm, a total of 12 frames per heart cycle was obtained.
For the first series, a Black Blood (BB) CINE FLASH sequence was additionally performed with 6 frames and a
120 ms time of inversion (TI), using optics and ECG signals. For each prospectively-gated acquisition series,
cardiac and respiratory periods as well as total scan duration were recorded.
For retrospective gating, the IntraGateFLASH method from Bruker was used. The read- and phase-dephase
gradients were separated from the slice-refocusing gradient in order to detect a half echo signal without phase
encoding. The navigator was then derived from the selectively excited slice. It is worth noting that this technique
is restricted to a single slice, since the navigators from neighboring slices cannot be combined. The imaging
parameters were TR/TE=10.3/4.2 ms; 1.1 mm slice thickness; 200 repetitions for 6 min 36 sec total scan time.
Other acquisition parameters were identical to those used in the FLASH method. CINE image reconstruction
was performed with one respiration frame and 12 heart frames to be consistent with prospective parameters.
Finally, the efficiency of the optical gating device was also assessed on mice liver using a dual cardio-respiratory
triggering. An axial fat suppressed (FS) multiple Spin-Echo (SE) sequence was used with the following
parameters: 30x30 mm
2
FOV; 0.7 mm slice thickness; TR=6000 ms; TE=20, 40 and 60 ms; 256x192 matrix, 24
slices using previously describe acquisition strategy [22, 23].
E. Image analysis
For heart imaging, a region of interest (ROI) corresponding to the left ventricular myocardium wall was drawn
manually on every single CINE image using CreaContour (in-house developed software). Mean signal intensity
and surface area in the ventricular cavity and in the myocardium wall were calculated. The signal-to-noise ratio
(SNR) in the left ventricular myocardium was then assessed by dividing the average signal intensity by the
standard deviation (SD) of the image background noise level in an out-subject region free of ghosting artifacts.
The contrast-to-noise ratio (CNR) value was computed as the SNR difference measurement between the
myocardium wall and the left ventricular cavity which corresponds to the end-diastolic phase with no flow
artifacts.
For each series, significant differences between SNR, CNR, scan time and physiological period values measured
with the different sensors or techniques were determined using a paired Student’s t test (Excel, Microsoft, WA,
USA). For liver imaging, the efficiency of the optical-based triggering signal was simply assessed based on a
visual analysis.
III. RESULTS
A. Optical device
The complex transfer function of the optical device was measured. Amplitude and phase were plotted against
frequency (Fig. 3a). The low and high corner frequencies were higher than 0.3 Hz and higher than 30 Hz
respectively. The device response to an ECG-like signal is slightly distorted but retains a similar aspect to the
driving signal (Fig. 3b).
However, a typical delay of several milliseconds was systematically observed between each wave of the
driving signal (DS) and the corresponding response signal (RS). For instance a time shift of 14 ms was measured
between the R-wave of RS and the corresponding component of DS. This induces a delay of about 10% of the
cardiac cycle duration between the thorax motion signal and the trigger unit’s output when the optical device is
used compared to ECG or pressure sensors. Moreover, the shape of the main waves is modified as the output
waveform is smoothed by the filter. The measured sensitivity of the device within the bandwidth was typically
25 mV output voltage per nW of input light.
B. Optical-based motion signal
Only a few seconds were necessary to install the optical probe correctly and visualize a signal with both
cardiac and respiratory components. The signal is comprised of distinct peaks representing respiratory and heart
motions respectively (Fig 4a). The largest peaks are attributed to breathing while the smaller oscillations to heart
motion. The electromagnetic perturbations induced by RF pulses and gradient switching did not affect the optical
signal. The latter was independent of RF flip angle pulse and sequence used. The signal amplitude was high
enough to perform a straightforward adjustment of the gating levels with good differentiation between cardiac
and respiratory signal amplitude. Signal amplitude variations due to different experimental conditions such as
animal size, animal hair as well as the fiber tip location on the thorax are compensated by the adjustable
