DEVELOPMENT OF A DYNAMIC TEST PHANTOM
FOR OPTICAL TOPOGRAPHY
Peck H. Koh, Clare E. Elwell, and David T. Delpy∗
Abstract: Optical topography (OT) is a near infrared spectroscopy (NIRS) technique that
provides spatial maps of haemodynamic and oxygenation changes. When developing,
testing and calibrating OT systems it is often necessary to use tissue simulating phantoms
that are capable of providing realistic changes in attenuation properties. We present a
novel dynamic tissue phantom that enables spatially and temporally varying tissue
properties to be reproduced in a controlled manner.
This new dynamic test phantom consists of a modified liquid crystal display (LCD)
(enabling flexible and rapid changes in attenuation across different regions of the
phantom) sandwiched between two layers of tissue simulating epoxy resin (providing
static and homogeneous optical absorption and scattering). By activating different pixels
in the liquid crystal display it is possible to produce highly localised and dynamic
changes in attenuation which can be used to simulate the changes associated with the
cerebral haemodynamic response to functional activation. The reproducibility of the
dynamic phantom will be described with examples of its use with an OT system.
1.
INTRODUCTION
The development and testing of most imaging systems requires the use of tissue
phantoms that are capable of providing realistic changes in attenuation properties. These
phantoms can be used to test design prototypes and evaluate instrument performance.
Many useful optical phantoms have been developed in recent years with a wide range of
attenuation properties that are characteristic of biological tissues. Optical phantoms for
NIRS studies were first developed for breast imaging studies1 and subsequent research
into different types of optical imaging techniques has led to the generation of a variety of
tissue phantoms. Many of these have focused on the design of regular-shaped objects
with specific attenuation properties including the use of biological molecules such as
∗
Department of Medical Physics and Bioengineering, University College London, Gower
Street, London WC1E 6BT, United Kingdom. pkoh@medphys.ucl.ac.uk
2
KOH ET AL
haemoglobin and melanin as absorbing components2 and the generation of hybrid
phantoms for multimodality imaging.3 Classical phantoms generally have static
properties i.e. absorption (µa) and reduced scattering coefficients (µs’) matching those of
the biological tissues4 that are not spatially variable.
NIR topography systems are increasingly being used to monitor the haemodynamic
response to functional activation via the resulting spatially and temporally varying
changes in cortical attenuation. Simulation of this varying physiological signal cannot be
carried out using the static phantom approach and there is a need for a dynamic phantom
that can reproduce these optical changes spatially and temporally in a controlled manner.
Design of a Dynamic Phantom using Liquid Crystal Display (LCD)
One approach to producing a phantom which enables rapid and spatially varying
changes in attenuation is to use the existing design of a multilayer resin phantom and to
insert an additional layer of material within it, the optical properties of which can be
altered easily. An LCD seems to fit the requirements for such an inserted layer since 1) it
relies on external light to provide the image and does not emit light itself and 2) it has a
layer of light-polarising liquid molecules whose orientation is electrically controlled
which allows its attenuation properties to be changed rapidly and with high spatial
specificity. This combination of static layers in which the optical properties are fixed
(representing skin/skull and white matter) and a dynamic layer in which the attenuation
properties can vary (representing cortical tissue) provides a reasonable simulation of the
type of physiological signal changes that the OT systems are required to monitor.
Figure 1a shows the configuration of the LCD dynamic phantom. The phantom
incorporates a top layer of epoxy resin with a realistic thickness (to represent the
attenuation due to extracerebral layers). The light then passes through the LCD display.
In the current study the size of LCD unit used allowed six laser sources and three
detectors to be configured to produce measurements over seven channels with a fixed
optode spacing of 30 mm (Figure 1b). The lower block phantom material (representing
the brain white matter) is large enough to ensure light is scattered back to the surface and
light loss to the boundary is small.
(a)
(b)
Figure 1. (a) Schematic of the dynamic phantom and (b) optode arrangement used in the current study.
DEVELOPMENT OF DYNAMIC TEST PHANTOM FOR OPTICAL TOPOGRAPHY
2.
3
METHODS
A QVGA (resolution: 320 x 240 pixels) transflective mode LCD (Nan Ya Plastics
Corporation) was used. The 2.2 mm-thick passive display (active area: 76 x 57 mm) uses
a Chip-on-Flex configuration. Each active pixel in the LCD has an area of 5.64 x 10-3
mm2. An external controller (CB-GT380, Amulet Technologies Limited) was selected to
drive and control the display. A HTML script describing the sequence of the animation is
programmed into the controller. All optical measurements were made using a ETG-100
OT system (Hitachi Medical Corporation, Japan).5 The results described are from the
light intensity measurements of the 780 nm-wavelength, expressed in arbitrary units.
The recipe used to construct the solid phantom has previously been published.4, 6 The
matrix material was constructed using epoxy resin (MY 753 Aeropia Chemical Supplies).
A concentrated dye solution (Pro Jet 900NP Zeneca Ltd.) as absorber and a Superwhite
polyester pigment (Alec Tiranti Ltd.) as the scatterer were mixed with the solution. The
resulting phantom has a µa of 0.01 mm-1 and µs’ of 1.0 mm-1 (at 800 nm) and the mean
cosine of scatter is about 0.5.4 The area of the epoxy resin block was 110 x 90 mm2 with
the base block having a thickness of 30 mm and the top block 5 mm. The absorption and
scattering coefficients of the static phantom block were measured using a time-resolved
optical system (MONSTIR).7 The test result showed that the variation between expected
and measured coefficients was less than 5 %.
3.
RESULTS
In OT it is often assumed that the measured intensity correlates with the size of the
activated attenuating region. However this will only operate over a limited range of sizes
and shapes since in a scattering medium light can take various routes between the source
and detector including paths which do not pass through the activated region. Initial
studies were therefore conducted to investigate (i) the size of attenuating region that can
be detected by OT, (ii) the effect of different region geometries and (iii) “crosstalk”
between different optodes with spatially varying attenuation regions.
Effect of Attenuation Region Size
A circular attenuation region was programmed on the display area positioned centrally
between source 1 and detector 3 (i.e. at the position of channel 4). The diameter of the
region was varied between 1 and 27 mm. To determine the contrast (i.e. intensity
difference between two different conditions) the detected intensities were normalised to
an “all-dark display” baseline condition where all the pixels were activated.
Figure 2 shows the normalised intensity as a function of region diameter. The results
show a generally linear relationship between detected intensity and attenuator size down
to a diameter of 5 mm. Below 5 mm, the contrast, which we have arbitrarily defined as a
change of less than 1 %, approached the system noise level.
4
KOH ET AL
Figure 2. Attenuation time course for varying attenuator sizes (indicated at top of figure). The alternate grey
and white areas indicate the period for each size variation and the normalised intensity changes (indicated at
bottom of figure) were calculated with reference to a baseline condition where the whole LCD was dark.
Effect of Attenuation Region Shape
Figure 3. Comparison of the effect of attenuation regions with different shapes.
DEVELOPMENT OF DYNAMIC TEST PHANTOM FOR OPTICAL TOPOGRAPHY
5
The LCD was programmed to produce a range of attenuation regions with equal
areas but different shapes, including squares, circles, rectangles with fixed width
(vertical) and rectangles with fixed height (horizontal). The regions were centred on
channel 4 and their areas were varied between 1 and 576 mm2. The intensity changes
were subsequently normalised to the previously described baseline condition. Figure 3
shows a plot of the normalised intensity changes (expressed in percentage difference) for
each of the attenuator shapes as a function of attenuator area. The result shows a
generally linear trend between intensity and region sizes, but identifies a significant
difference between the two rectangular shapes with same areas. The variation is due to
the orientation of the optode pair relative to the region. The results suggest that along
with size variation, differences in shape and orientation of the activated region can affect
the OT signal.
Effect of Attenuator Position on Crosstalk between Channels
A circular attenuation region with a diameter of 27 mm was programmed on the
LCD to move horizontally in sequence along the midline from channels 1-2, 3-5 and 6-7.
The amount of “crosstalk” was defined as a residual change detected in the neighbouring
channels. Figure 4 shows the intensity changes on all 7 channels as the 27 mm diameter
attenuator was moved across the phantom. 10 episodes of crosstalk were identified
(highlighted as circles) during this test. Subsequently, the attenuator region size was
gradually reduced until the crosstalk effect was considered to be negligible (less than 1 %
change in intensity). This occurred when the diameter of the attenuator region was less
than 21.5 mm.
Figure 4. Normalised intensity changes for all 7 channels with the 27 mm attenuator region. 10 “crosstalk”
effects are circled.
6
4.
KOH ET AL
DISCUSSION
A new dynamic phantom using a modified LCD has been developed as a tool for
evaluating and optimising OT instruments. By placing the LCD between two blocks of
static phantom materials which provide well-characterised optical properties, a more
realistic simulation of the spatially and temporally varying attenuation changes seen
during cortical functional activation can be achieved. The effects of attenuator size, shape
and orientation on measured intensity have been described. In addition, evidence of
“crosstalk” between different optode pairs has been identified. This initial assessment
suggests that, for the phantom geometry described in this study, the OT system was able
to provide reasonable spatial differentiation with various attenuator shapes and sizes.
There are several issues which were not addressed in this study; the attenuation
effects associated with the LCD particularly the loss of light due to the polarizers and
effect of the clear glass on the LCD have not been quantified. While the contribution due
to the difference in refractive indices between the epoxy resin (1.56 at 800 nm) and the
LCD glass (~1.5) is considered to be negligible, coupling between the LCD and static
phantom layer has not been investigated. The current setup can also only simulate
attenuation changes in two dimensions, when in reality light attenuation due to
physiological changes in tissue occurs in three dimensions.
In most cases optical topography is used to measure the spatial changes in
attenuation arising from changes in chromophore concentration and as such absolute
quantification of the attenuation properties is not required. Since the dynamic phantom is
also intended to be used as a calibration tool for system, the actual photon path (which in
turn affects the sampling area) will not affect the measurement as long as the variation
remain fairly consistent throughout the measurement volume.
5.
ACKNOWLEDGEMENTS
This work was funded with grants from EPSRC/MRC, (GR/N14248/01) and in
collaboration with Hitachi Advanced Research Laboratory, Japan.
6.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
J.Linford, S.Shalev, J.Bews, R.Brown, H.Schipper, Development of a tissue-equivalent phantom for
diaphanography, Medical Physics, 13(6): 869-875 (1986).
A.M.De Grand, S.J.Lomnes, D.S.Lee, M.Pietrzykowski, S.Ohnishi, T.G.Morgan, A.Gogbashian,
R.G.Laurence, J.V.Frangioni, Tissue-like phantoms for near-infrared fluorescence imaging system
assessment and the training of surgeons, Journal of Biomedical Optics, 11(1): 014007 (2006).
W.D.D'Souza, E.L.Madsen, O.Unal, K.K.Vigen, G.R.Frank, B.R.Thomadsen, Tissue mimicking
materials for a multi-imaging modality prostate phantom, Medical Physics, 28(4): 688-700 (2001).
M.Firbank and D.T.Delpy, A design for a stable and reproducible phantom for use in near infra-red
imaging and spectroscopy, Physics in Medicine and Biology, 38: 847-853 (1993).
Y.Yamashita, A.Maki, H.Koizumi, Measurement system for noninvasive dynamic optical topography,
Journal of Biomedical Optics, 4(4): 414-417 (1999).
M.Firbank, M.Oda, D.T.Delpy, An improved design for a stable and reproducible phantom material for
use in near-infrared spectroscopy and imaging, Physics in Medicine and Biology, 40: 955-961
(1995).
F.E.W.Schmidt, M.E.Fry, E.Hillman, J.C.Hebden, D.T.Delpy, A 32-channel time-resolved instrument
for medical optical tomography, Review of Scientific Instruments, 71(1): 256-265 (2000).