Wavelength and Space Code Division Multiplexing

in Optical Tomography
L. Ascari1,3, G. Berrettini2, M. Giacalone2, L. Potì2
1 Scuola Superiore Sant’Anna, CEIICP, Pisa, 56125, Italy
2 CNIT, Pisa, 56127, Italy
3 HENESIS srl, Parma, 43123, Italy

email: luca.ascari@sssup.it

Summary
This paper introduces the use of coding techniques, applicable to time-resolved
diffuse optical topography and tomography imaging systems, for increasing
performances in terms of signal to noise ratio and acquisition speed. Parallel
acquisition of several wavelengths and locations is achievable.

Introduction
Optical Tomography aims at reconstructing one or more spatially variable coefficients
in a volume from measurements of propagation of near infrared light fluxes at its
surface. The propagation of near infrared photons in biological media has a diffusive
nature, as photons are absorbed and massively scattered by tissues, and coefficients
are reconstructed by post-processing algorithms. Time resolved techniques allow for
concurrent estimation of both absorption and scattering coefficients extracted from
the temporal point spread functions (TPSF). Previously introduced hardware and
software technologies are commonly referred to with the name of Time-Domain
Diffuse Optical Tomography (TD-DOT) [1]. Most of TD-DOT techniques use ultrashort
light pulses for tissue illumination and Time-Correlated single photon counters (TC-
SPC) for detection. Although these systems offer high dynamic range, sensitivity and
linearity, they are bulky, present a high cost and require absolute darkness during the
long acquisition. A spread spectrum approach [2] based on continuous wave (CW)
source modulation by a pseudo-random bit sequence (PRBS) allows overcoming
these limitations by combining the advantages of time-resolved systems with those of
low cost of components, lower sensitivity to ambient light, shorter acquisition time.
Nevertheless only one wavelength and one emitting location at a time can be
switched on, thus making the full acquisition from an array of sources and detectors a
time consuming serial process. This paper presents a novel architecture which
exploits the code division multiplexing (CDM) technique, commonly used in
telecommunications, to overcome also these last limitations. The architecture of the
core element of the tomographic system, the encoder, is presented (Fig. 1), together
with first simulations and experimental results (Fig. 2).

Methods, Discussion and Conclusions

Fig. 1 contains the functional description of the encoder in the configuration with four
wavelengths (λ1 - λ4), three launch sites (S1 - S3) and one detector. Each light source
is a double-encoded cheap CW distributed feedback (DFB) laser diode: the first set
of words (C1 - C4) encodes the wavelength, the second (C5 - C7) the emission
position. The emission power is dynamically and continuously equalized (this is a
general requirement of CDM system). All the diodes are always on (no scanning is
necessary). The detector (the square in the figure) can be either an avalanche
photodiode (APD) or a SPC, and reconstructs the global signal, containing
 all the wavelengths, from all the sources, at all
their multiple paths. The correlation receiver
operates a double decoding to reconstruct the
single TPSFs: for instance, contribution from
source S1 at λ1 is obtained by cascade
correlating the received signal with C5 and
then C1. If C1-C7 are orthogonal and have
dirac-like autocorrelation the detection of
delayed replicas of each channel is
possible and TPSF extracted. A Matlab-
based simulator was conceived to compare the
performances of PRBS, Gold and Walsh codes
with reference to the level of multipath, the
variance of the photon propagation time, the
spectrum spreading factor, the length of the codes. Then a preliminary experimental
activity was focused on the detection of two optical signals and related replicas
generated by a fading-affected channel. In the specific case, the two signals are
1541 nm and 1547 nm CWs generated by DFB laser and directly modulated with a
proper 27-1-long PRBS at 1.27 MHz. Each symbol is coded by a 15 chip-long Gold
short sequence at 19.05 MHZ. The PRBS pattern period is 100 µs, and two replicas
were sent on the fibre. Coded signals are coupled into a single fibre. The multipath is
obtained by splitting the optical signal and by using a 5900 meter-long fibre spool in
one branch, introducing 29.5 µs delay. The detector is a New Focus IR DC 125 MHz-
bandwidth low-noise avalanche photodiode (APD) and an Agilent 5464A 500 MHz
bandwidth 2 GSa/s oscilloscope is used to visualize the electrical signal. A LabView
software platform was developed to control the instruments parameters and to
acquire data streams by the oscilloscope traces, while decoding was performed by
means of a correlation receiver implemented in Matlab.
Fig.2 summarizes simulated (left) and experimental (right) results; the two top rows
are correlation signals, while the bottom ones are the reconstructed TPSF. The
separation between the higher peaks is exactly 100 µs, and 29.5 µs separate the
main from the delayed peaks, as expected. The delay between the two channels is
around 3 ns, as expected from the different optical path length. Noise level (standard
deviation of the correlation) is higher in real than in simulated conditions (0.010 vs
0.003); SNR on real data is 17dB. The delayed vs main signal amplitude is 55% in
simulation, but only 34% in real data: reasons are under investigation.
Used frequencies are not in the NIR
biological window, and fibre-induced
multipath is not what will be seen in
biological tissues: nevertheless, the aim
of these experiments was to assess the
principle; further investigation is clearly
needed, encouraged by these
promising preliminary results.
References
[1] D. A. Boas et al, “Imaging the body with diffuse optical
tomography,” Ieee Signal Processing Magazine, vol. 18, no.
6 , Nov. 2001. [2] W. Mo and N. Chen, “Design of an
advanced time-domain diffuse optical tomography system,”
Selected Topics in Quantum Electronics, IEEE Journal of,
vol. PP, no. 99, 2009.