Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Mueller Matrix Measurement Techniques for

various Optical Elements & Biological Tissues
Optics Lab Report By

Harsh Purwar (07MS – 76)

4th Year, Int. M.S. (Physics)
Indian Institute of Science Education and Research, Kolkata

Abstract

It has been observed that in recent years, researchers have given much emphasis on exploring
on various tissues through polarized light imaging technique as a potential diagnostic tool for
detecting any kind of abnormality particularly chronic diseases such as Cancer, Tumor, etc. As
we are aware of the fact that tissues depolarize a large fraction of incident light so that the
Mueller calculus lends itself well to these applications. Cancerous tissues were well
discriminated from normal tissues using Mueller imaging. Here I present a detailed analysis
about the various methods that can be adopted for measuring/calculating Mueller matrix for
various optical elements (such as polarizers, mirror, quarter wave plates etc.) and biological
samples (cancerous human cervical tissues). The analyses presented in this report includes
decomposition of the Mueller matrix constructed from the 16 polarization state measurements
(images and spectra) of various samples using polarization state generators and analyzers into
three optical parameters namely depolarization, di-attenuation and linear retardance.

Keywords: Mueller matrix, spectral measurement, depolarization, di-attenuation, birefringence, tissue

optics, polarization imaging, retardance, Mueller decomposition, turbid polarimetry, Stokes vector,
polarization, multiple scattering.

Introduction

Stokes Vector and Mueller Matrix

Stokes vector is a 4 element column matrix that describes the polarization state of electromagnetic
radiations. The elements of a Stokes vector are called Stokes parameters. These were defined by George
Gabriel Stokes in 1852. In terms of the components of electric field ( and ), the stokes parameters are
given by (1):
〈 〉
〈 〉
〈 〉
〈 〉
Here are the 4 elements of Stokes vector or Stokes parameters.
And the intensity of light by

1
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

The polarization state of the scattered light in the far zone is described by the Stokes vector connected
with the Stokes vector of the incident light. If is the initial incident Stokes vector and is the Stokes
vector describing the scattered light then we have,

where is the normalized scattering matrix or Mueller matrix.

Polarization States

Polarization states are various orientations of a polarizer or analyzer. There can be infinitely many
orientations and so infinitely many states. These all experiments are designed to use only seven such states
listed below. Important point to remember is the relative differences between the states. For example
horizontal polarization state (H) should be more or less from the vertical polarization state ( ). Same
goes for +45 ( ) and -45 ( ).

Mueller Matrix

Out of the many ways of constructing Mueller matrix or rather measuring Mueller matrix experimentally
for any given optical element one of the most standard method and adopted by most, is the one described
using the concept of polarization state generators and analyzers.
For a particular wavelength of light, Mueller matrix was generated from the four incident polarization
states (linear polarization at angles of , , from the horizontal and right circular polarization) and
recorded the intensity of the light transmitted through sample after it passed through the suitably oriented
analyzers (linear polarization at angles of , , from the horizontal and right circular polarization).
Mathematically we use PSG and PSA matrix respectively as,

( ) ( )

intensity measurement matrix is given as,

is the Mueller matrix of the sample, which can also be written as vector,

where is a matrix given as Kroneker product of PSA with transpose of PSG,

This kind of construction allows simpler calibration for non-ideal components since one replace the ideal
stokes vector in each column of PSG or each row of PSA by measured stokes vectors that may deviate from
ideal value which is not possible in direct construction of Mueller matrix from other intensity
measurements. The errors arising due to the finite extinction ratios of the linear polarizer and analyzer was
incorporated using the generated matrix corresponding to these polarizers.
For a white incident light, the set of four incident polarization states were generated by rotating fast axis of
quarter wave plate (QWP) with respect to the pass axis of first polarizer (Figure 3). For each input
polarization state, intensities at four different analyzer states are measured using suitably oriented QWP
and polarizer at the output. Stokes vector of light incident on sample after passing through a fixed
polarizer and QWP (for a given orientation angle, ), is given as,

2
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

( )
( ) ( ) ( ) ( )
( )

( )
( )

where, , , , and is the linear retardance which for a

QWP is ⁄ .
Above equation represents the general Stokes vector of the incident light. Now, changing will change its
polarization state. We chose four different orientation angles ( , , and ) of QWP and hence PSG
comes out as,

( )
( ) ( ) ( ) ( )

Similarly the polarization state of the detected light, analyzed by PSA optics will depend upon the
orientation angle of QWP . In Mueller matrix measurements only intensity is recorded which is given by
the first element of the Stokes vector. In our case this is decided by the first row of the matrix formed by
the multiplication of Mueller matrix of QWP (for a given ) followed by the Mueller matrix of polarizer
(kept cross with respect to which is at horizontal position). By changing orientation angles of QWP
( , , and ) four different polarizations of the scattered light can be analyzed. Hence PSA can be
written as,
( ) ( )
( ) ( )
( ) ( )
( ) ( )
( )

And then rest of the algebra remains same as mentioned above for the previous case.

Polarization Images

In one of the following experiments, polarization images are the images taken by the CCD for various
polarization states like , , , etc. These polarization images may also be called as the intensity images
for various states. Basically these intensity images are matrix, each element storing the value
of intensity (in arbitrary units) falling at that particular location on the CCD’s chip. This intensity can have
values ranging from 0 to 255. While processing the images and decomposing the matrix using a simple
MATLAB script (Appendix section) the value of each parameter is calculated for all pixels of the images.

3
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Optical Parameters

 Depolarization:
A process which couples polarized light into un-polarized light. Depolarization is intrinsically
associated with scattering and with di-attenuation and retardance which vary in space, time, and/or
wavelength. Examples of depolarizers include tissues, polystyrene microspheres, phantoms etc.
 Di-attenuation:
The property of an optical element or system whereby the intensity transmittance of the exiting
beam depends on the polarization state of the incident beam. The intensity transmittance is a
maximum (Tmax) for one incident state, and a minimum (Tmin) for the orthogonal state. The di-
attenuation is defined as (Tmax - Tmin) / (Tmax + Tmin).
Any homogeneous polarization element which displays significant di-attenuation and minimal
retardance is called a di-attenuator. Polarizers have a di-attenuation close to one, but nearly all
optical interfaces are weak di-attenuators. Examples of di-attenuators include the following:
polarizers and di-chroic materials, as well as metal and dielectric interfaces with reflection and
transmission differences described by Fresnel equations; thin films (homogeneous and isotropic);
and diffraction gratings.
 Polarizance:
The property of an optical element or system whereby un-polarized light is transformed into
polarized light. The polarizance is described by its magnitude (equal to the degree of polarization of
light exiting the system when un-polarized light is input) and the Stokes vector of the output light.
 Retardance:
A polarization-dependent phase change associated with a polarization element or system. The
phase (optical path length) of the output beam depends upon the polarization state of the input
beam. The transmitted phase is a maximum for one eigen polarization, and a minimum for the
other eigen polarization. Other states show polarization coupling and an intermediate phase.
 B-irefringence:
A material property, the retardance associated with propagation through an anisotropic medium.
For each propagation direction within a bi-refringent medium there are two modes of propagation
with different refractive indices and . The bi-refringence is given by, | |.

Decomposition Scheme for Mueller matrix

An arbitrary Mueller matrix can be decomposed into three basic optical parameters namely
decomposition, retardance and di-attenuation discussed in the latter half of this section. It has been shown
that any Mueller matrix can be expressed as a product of three matrices called depolarizer, di-attenuator
and retarder (2). So we have,
( )

The three di-attenuation components ⁄ , ⁄ and provide its complete description. The
di-attenuation of a Mueller matrix described above is given by,
√ ( )

Similarly di-attenuation vector is given by,

4
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

⃗ ( ) ( ) ( )

And diattenuation matrix is given by,

⃗
[ ] ( )
⃗
where is given by,

√ ( ) ( √ )⃗ ⃗ ( )

So, we are now left with,

( )
( )

Let and be the eigenvalues of ( ) . So from Eq. – 7, has √ √ and √ as

eigenvalues. After calculating the eigenvalues of ( ) we evaluate the following expression,
[ ( ) (√ √ √ ) ]
[(√ √ √ ) ( ) √ ] ( )
If the determinant of is negative then minus sign is applied otherwise positive. Now to get from
we first calculate polarizance vector and polarizance matrix using,

⃗ ( ) ( )

⃗ ⃗
⃗ ( )

Now we write as a matrix in terms of ⃗ and as below,

⃗
[ ] ( )
⃗

The value of depolarization power or simple depolarization is also calculated and is shown as a figure
(later) using,
| ( ) |
( )

Now again we pre-multiply (Eq. – 6) by to get,

( )

The value of retardance was calculated as,

( )
[ ] ( )

Spatial Light Modulators:

A spatial light modulator (SLM) is an object that imposes some form of spatially – varying modulation on a
beam of light. A simple example is an overhead projector transparency. Usually when the phrase SLM is

5
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

used, it means that the transparency can be controlled by a computer. In the 1980s, large SLMs were
placed on overhead projectors to project computer monitor contents to the screen. Since then more
modern projectors have been developed where the SLM is built inside the projector. These are commonly
used in meetings of all kinds for presentations.
Usually, an SLM modulates the intensity of the light beam. However, it is also possible to produce devices
that modulate the phase of the beam or both the intensity and the phase simultaneously.
SLMs are used extensively in holographic data storage setups to encode information into a laser beam in
exactly the same way as a transparency does for an overhead projector. They can also be used as part of a
holographic display technology.
SLMs have been used as a component in optical computing. They also often find application in holographic
optical tweezers.

Optically addressed spatial light modulator (OASLM):

The image on an optically addressed spatial light modulator, also known as a light valve, is created and
changed by shining light encoded with an image on its front or back surface. A photo-sensor allows the
OASLM to sense the brightness of each pixel and replicate the image using liquid crystals. As long as the
OASLM is powered, the image is retained even after the light is extinguished. An electrical signal is used to
clear the whole OASLM at once.
They are often used as the second stage of a very-high-resolution display, such as one for a computer-
generated holographic display. In a process called active tiling, images displayed on an EASLM are
sequentially transferred to different parts on an OASLM, before the whole image on the OASLM is
presented to the viewer. As EASLMs can run as fast as 2500 frames per second, it is possible to tile around
100 copies of the image on the EASLM onto an OASLM while still displaying full-motion video on the
OASLM. This potentially gives images with resolutions of above 100 megapixels.

Experimental Setup

For a Fixed Wavelength:

Figure 1: Schematic Diagram of the Experimental Setup in transmission mode geometry designed for a fixed
wavelength of light.

Key:

 Helium (He) Neon (Ne) LASER (633 nm 12.0 mW)

 Neutral Density (ND) Filter manufactured by ThorLabs was used.
 Polarizers, and manufactured by ThorLabs were used.
 Quarter Wave Plates, and again manufactured by ThorLabs were used.

6
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

 The rotational mounts for the polarizers and QWPs were specially designed for precise angular
measurements.
 Sample is an optical element whose Mueller matrix we are interested in finding out.
 A collecting lens can be used for better results in case the sample is highly scattering or has high
absorbance in between the sample and QWP .
 Final measurement was intensity based done using a charge coupled device or a CCD camera
manufactured by ThorLabs. The intensity measurement range of the CCD was from 0 – 255 in
arbitrary units (0 for darkest pixel and 255 for the brightest).
 Another convex lens of appropriate focal length could be used to focus the light from the polarizer
to the CCD chip in case the beam diverges too much.
 The CCD was connected to the computer and the intensity image (profile) as seen by the CCD chip
was recorded using the software provided along with the device.

Figure 2: Experimental Setup for white light in back scattering mode.

7
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

For White Light (Xenon Lamp):

Figure 3: Schematic Diagram of the Experimental Setup in the back-scattering mode geometry designed for white
light from Xenon Lamp.

Key:

 The above shown apparatus was mounted on a rail whose ends were fixed on a rotational stage.
This helps in aligning the optical elements and back-scattering angle can be measured precisely.
 Xenon Lamp (50 W)
 Polarizers and are wide band transmission polarizers manufactured by ThorLabs.
 Similarly quarter wave plates and also have a wide band transmission again manufactured by
ThorLabs.
 Any other optical element required in the experiment should also have a wide band transmission
spectrum like focusing and collecting lenses, optical fibers etc.
 The light was focused on one of the ends of an optical fiber which was then connected to a
spectrometer.
 The measurement was done in the form of a spectrum (Intensity versus Wavelength curve)
recorded using the software supplied along with the spectrometer.

Equipment Calibration

Before starting the experiment according to the above setup, we must assign rather choose 4 states of
input and output polarizations of light commonly known as Polarization State Generator (PSG) and
Polarization State Analyzer (PSA). It has been shown that one can find out the Mueller matrix using any 4
arbitrary states as far as the determinant of the W-matrix is non-zero. We choose the following standard

8
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

four states which gives the determinant of the W-matrix maximum (far away from zero) and also this
arrangement suits our setup for a particular wavelength.

For a fixed Wavelength:

Remove and . Let us call to be the horizontal, state of the polarizer . The corresponding cross-
state (vertical) of the polarizer can be found using one of the fundamental laws in optics known as
Malus’s Law. The polarizer is rotated to obtain minimum intensity on the CCD or a photo diode. The
angle for is recorded and this state is called vertical, . Similarly state for would be and
corresponding state could be noted for , would be vertical ( ) for and for can be found.
Now to find the of the QWP set to state and to state . Now find an angle of for which
on rotating the polarizer the intensity profile remains almost constant. This angle along with state of
corresponds to the right circularly polarized light abbreviated as . Same is done for the QWP .

Following table shows the angles corresponding to the four states for , , and .

For Polarizer :
Horizontal ( )
Vertical ( )
+45 ( )

For Polarizer :
Horizontal ( )
Vertical ( )
+45 ( )

For QWP :
+45

For QWP :
+45

For White Light (Spectra Measurement):

The apparatus for the white light is as shown in Figure 3. The QWP was mounted on a computer
controlled rotational mount. The polarizers and QWPs can easily be calibrated as mentioned above. Here
we have chosen four states of QWP’s and the polarizers were fixed in a particular state. was fixed at
horizontal and was fixed at vertical. The four chosen angles for the QWPs were optimized so as to get a
maximum possible determinant of the W-matrix. The optimization code (written in MATLAB) for the same
is attached in the Appendix at the end of this report. The four chosen angles were , , and
. The corresponding determinant of the W-matrix was computed to be 18.23.

9
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Procedure

Among many ways of experimentally measuring Mueller matrix for an optical element the two adopted
methods with followed protocol are as follows.

For a Fixed Wavelength:

After cleaning and aligning the optical equipment as mentioned above in the Experimental Setup section
data images were recorded as detailed below.

 Before recording the data optical alignment was checked ones again.
 The optical element whose Mueller matrix is to be measured was mounted on the sample stand
and it was made certain that the LASER beam passes though the approximate center of the sample.
 Both the two polarizers, and were set to angles corresponding to the state (horizontal).
Both the QWP’s were removed temporarily.
 The image as seen through the charged coupled device (CCD) on the computer screen was saved
after adjusting the integration time so that the device is not saturated. The integration time was
not changed after that.
 Similarly by rotating the two polarizers to proper angles known by calibrating each of them all other
8 states ( , , , , , , , ) were recorded.
 The QWP’s and were then introduced and following the same procedure rest of the 7 states
( , , , , , , ) were recorded.

For White Light:

A similar method as described above was followed for recording the spectra of various input and output
states. Briefly the adopted method is described below.

 Before recording the spectra optical alignment was checked ones again.
 The optical element whose Mueller matrix is to be measured was mounted on the sample stand
and it was made certain that the LASER beam passes though the approximate center of the sample.
 The two polarizers, and were set to angles corresponding to the state (horizontal) and
(vertical) respectively.
 The two quarter wave plates, and were rotated to the above calibrated angles and the
corresponding spectrum was recorded for each state.

10
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Observations

Recorded Images for various Input & Output States

HH HP HR HV

PH PP PR PV

RH RP RR RV

VH VP VR VV

For Blank (No Sample)

The fringe pattern visible in almost all the intensity images above is mainly a side effect of the use of a
circular pin-hole (or aperture) after beam broadening.

11
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Recorded Spectrum for Various Input & Output States

Following is a spectrum for one of the states with the QWP aligned at angles and for a
biological sample (human tissue with cervical cancer) mounted on a microscope slide.

Figure 4: Spectrum of white light incident on a human tissue sample

for the state with , , and .

Data Analysis

Analysis of data recorded in the form of images

 Each image for blank (no sample) and QWP (at ) as a sample was imported in MATLAB as a
matrix and was summed over all the pixels or elements so as to obtain an overall average intensity
corresponding to each state.
 The PSA and PSG were constructed for an ideal case and the corresponding W-matrix was
calculated.

( ) ( )

And,

 The Mueller column vector ( ) was then formed for blank as well as for the sample (QWP)
for the measured states.
( )

12
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

 Sample Mueller vector ( ) was then computed using the relation,

 This was then transformed into a matrix to obtain the Mueller matrix for blank and sample
(without blank correction).
 Then the corrected Mueller matrix for the QWP sample was obtained by,
( )

Results and Conclusions

The Mueller matrix obtained for the blank (no sample) is as follows,

( )

Ideally the Mueller matrix for blank should be identity but as can it seen it deviates from that. This is
mainly due to other non-ideal optical elements in the PSA and PSG optics. Using the above matrix blank
correction was done and the corrected Mueller matrix for the QWP at an angle of was calculated,

( )

Spectral Measurements

The spectral measurement of Mueller matrix was done for a QWP at an angle of and then this matrix
was decomposed to give depolarization, di-attenuation and linear retardance as a function of wavelength.
Following plots (Figure 5, Figure 6 & Figure 7) show these properties for a QWP in transmission mode
geometry.

Figure 5: Depolarization versus wavelength for a QWP in transmission mode geometry.

13
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Figure 6: Di-attenuation versus wavelength for a QWP in transmission mode geometry.

Figure 7: Linear retardance versus wavelength for a QWP in transmission mode geometry.

14
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

The same was done for a polished mirror as well but as is obvious in the back-scattering mode. Following
(Figure 8, Figure 9 & Figure 10) are the plots.

Figure 8: Depolarization versus wavelength for a polished mirror in the back-scattering mode.

Figure 9: Di-attenuation versus wavelength for a polished mirror in the back-scattering mode.

15
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Figure 10: Linear retardance versus wavelength for a polished mirror in the back-scattering mode.

Linear retardance for the mirror, at all wavelengths of the incident light, should be closed to zero but it
seems from the above figure that it is close to . This is due to the reflection of the light from the surface
of the mirror.
Apart from this the reason for periodicity in the results (in all previous curves) is not very clear at this
moment. It might be due to the misalignment of other optics involved in the experimental setup, or could
even be due to the imperfections in the mirror itself. The exact reason could not be explained.

16
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Biological Samples

A very similar measurement was done for the biological tissues (human cervical cancer tissues) for three
different grades of cancer. The beam was focused on the stroma region of the tissues. Obtained spectra
was analyzed using the code attached at the end in the Appendix section of this report. This was done in
both transmission and back-scattering mode geometries. Moreover, instead of doing it at zero degree we
preferred to do at in transmission mode so as to collect most of the scattered light from the tissue.
Similarly in the back-scattering mode the angle between the incoming beam and the collected light beam
was . Following figure (Figure 11) shows variation of each element of Mueller matrix with wavelength
measured for the human cervical cancer tissue (Slide Code: 8413) in the back-scattering mode with angular
separation between the incident and the detected light being .

Figure 11: Mueller elements versus wavelength for human cervical cancerous tissue Grade 1 (Slide Code: 8413).

Each of the above Mueller elements plays an important role in the diagnostic technique for cancer.
However as described above, the three more physical parameters namely depolarization, di-attenuation
and linear retardance would be used as a tool here for the basic diagnostic of cancer or more precisely as
parameters to distinguish between various grades/stages of cancer in these biological samples.

17
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Following are the plots of depolarization, di-attenuation and linear retardance for these 3 grades of
cancerous tissues.

Depolarization Plots for Biological Samples

Figure 12: Depolarization versus wavelength for human cervical cancerous tissue, Grade 2 (Slide Code: 7695).

Figure 13: Depolarization versus wavelength for human cervical cancerous tissue, Grade 1 (Slide Code: 8413).

18
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Figure 14: Depolarization versus wavelength for human cervical cancerous tissue, Grade 3 (Slide Code: 8666).

Di-Attenuation Plots for Biological Samples

Figure 15: Di-attenuation versus wavelength for human cervical cancerous tissue, Grade 2 (Slide Code: 7695).

19
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Figure 16: Di-attenuation versus wavelength for human cervical cancerous tissue, Grade 1 (Slide Code: 8413).

Figure 17: Di-attenuation versus wavelength for human cervical cancerous tissue, Grade 3 (Slide Code: 8666).

20
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Linear Retardance Plots for Biological Samples

In the following plots of linear retardance versus wavelength for the three tissue samples it seems that
experiments done in two different geometries yield very different results. This is not true in fact. The value
of retardance for the back-scattering geometry ( ) at each wavelength is as clear from the graph is
close to which is due to the reflection of the incident light. If we subtract from these values then the
results in both the cases match closely.

Figure 18: Linear retardance versus wavelength for human cervical cancerous tissue, Grade 2 (Slide Code: 7695).

Figure 19: Linear retardance versus wavelength for human cervical cancerous tissue, Grade 1 (Slide Code: 8413).

21
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Figure 20: Linear retardance versus wavelength for human cervical cancerous tissue, Grade 3 (Slide Code: 8666).

From the above figures (Figure 12 - Figure 20), one cannot precisely distinguish between various stages of
cancers as there are a lot of errors associated with the measurements. Also one has to do the experiment
in more idealized conditions for more number of tissue samples and with proper alignment. Also a more
thorough analysis of the data is required so as to account for any associated random errors etc. But
certainly this diagnostic technique could lead to a major breakthrough in the medical science.

Spatial Light Modulator (SLM)

The spectral measurement for the construction of Mueller matrix was also done with SLM as a sample so
as to study the basic optical properties of it. The following gray scale (default) was set with the help of
configuration software supplied along with it and the spectral measurements were done just as mentioned
before in the Procedure section.
Gray scale Details:
Contrast: 195 (Min: 0, Max: 255)
Brightness: 100 (Min: 0, Max: 255)

22
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Following are the decomposition plots for the spatial light modulator with above specified gray scale.

Figure 21: Depolarization versus wavelength for spatial light modulator at two angles (0° and 5°) in transmission
mode.

Figure 22: Di-attenuation versus wavelength for spatial light modulator at two angles ( and ) in transmission
mode.

23
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Figure 23: Linear retardance versus wavelength for spatial light modulator at two angles ( and ) in
transmission mode.

Sources of Errors

Out of the many possible sources of errors in the above described experiments following is a description of
a few major sources which could have affected the preceding results.

 Major contribution to the fluctuations in all most all of the above graphs comes from the
fluctuating initial (LASER/Lamp) intensity of light.
 Apart from this, fluctuations in case of biological samples are due to the fact that these samples
have complex structures/patterns, which even in the normal case would give a background
fluctuation in almost all measured optical parameters.
 Optical misalignment may lead to many spurious errors in the measurements.
 All the optical elements, to be used, must be thoroughly cleaned with acetone or lens cleansing
liquid/paper.
 Other deviations from the standard or expected results may be due to the involvement of other
non-ideal optics.

Acknowledgement

I thank Dr. Nirmalya Ghosh for his humble guidance, help and support throughout the experiment period. I
would also like to thank Jalpa Soni and my group partners, Satish Kumar, Amit Anand and Irfan Raza for
helping me in carrying out these experiments and analyzing the data.

24
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Works Cited

1. Tuchin, Valery. Optical properties of tissues with strong (multiple) scattering. Tissue Optics: Light
Scattering Methods and Instruments for Medical Diagnosis. Second. 1, pp. 3-108.
2. Interpretation of Mueller matrices based on polar decomposition. Lu, Shih-Yau and Chipman, Russell A.
5, Alabama : Optical Society of America, May 1996, Opt. Soc. Am. A, Vol. 13, pp. 1106-1113.
3. Wikipedia. Mueller Calculus. [Online] Wikipedia.org. http://en.wikipedia.org/wiki/Mueller_calculus.
4. Gupta, Vivek. Mueller matrix in optical imaging for cervical cancer detection. Department of Physics,
Indian Institute of Technology. Kanpur : s.n., 2009. pp. 1-6.
5. Reference Page. Axometrics Web Site. [Online] http://www.axometrics.com/reference.htm.
6. Ghosh, Nirmalya. Handbook of Photonics for Biomedical Science. s.l. : Unpublished. pp. 253-282.
7. General and self-consistent method for the calibration of polarization modulators, polarimeters, and
Mueller-matrix ellipsometers. Compain, Eric, Poirier, Stephane and Drevillon, Bernard. 16, June 1, 1999,
Applied Optics, Vol. 38, pp. 3490-3502.
8. Saleh, B E.A. and Teich, M. C. Fundamentals of Photonics. s.l. : John Wiley & Sons, Inc., 2007. ISBN:
9780471358329.
9. UTILIZATION OF MUELLER MATRIX FORMALISM TO OBTAIN OPTICAL TARGETS DEPOLARIZATION AND
POLARIZATION PROPERTIES. Roy-Brehonnet, F. Le and Jeune, B. Le. 2, s.l. : Elsevier Science Ltd., 1997,
Progress in Quantum Electronics, Vol. 21, pp. 109-151.
10. Purwar, Harsh. Mueller Imaging: An approach to Detect Abnormality in Human Brain Tissues.
Department of Physical Sciences, Indian Institute of Science Education & Research. Kolkata : Unpublished,
2009. Summer Project Report.

25
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Appendix

Code1:

For maximizing the determinant of W-matrix by choosing appropriate states of QWPs given the two
polarizers and are at and states respectively.

1 function fastest()
2 del=pi/2;
3 ang=0:2:180;
4 ang=ang/180*pi;
5 i=0;
6 [m,n]=size(ang);
7 for a=1:n
8 disp(a);
9 for b=a+1:n
10 for c=b+1:n
11 for d=c+1:n
12 PSG(:,1)=PSGG(ang(1,a),del);
13 PSG(:,2)=PSGG(ang(1,b),del);
14 PSG(:,3)=PSGG(ang(1,c),del);
15 PSG(:,4)=PSGG(ang(1,d),del);
16 PSA(1,:)=PSAG(ang(1,a),del);
17 PSA(2,:)=PSAG(ang(1,b),del);
18 PSA(3,:)=PSAG(ang(1,c),del);
19 PSA(4,:)=PSAG(ang(1,d),del);
20 W=kron(PSA,PSG');
21 DW=det(W);
22 if(DW>23.5)
23 i=i+1;
24 detW(i,:)=[ang(1,a)/pi*180,ang(1,b)/pi*180,ang(1,c)/pi*180,ang(1,d)/pi*180,DW];
25 end
26 clear W
27 end
28 end
29 end
30 end
31 save 'DetW.txt' detW -ascii;
32 % detW
33 end
34
35 function y1 = PSGG(t,del)
36 c2t=cos(2*t);
37 s2t=sin(2*t);
38 sl=sin(del);
39 cl=cos(del);
40 y1=[1;(c2t^2)+((s2t^2)*cl);c2t*s2t*(1-cl);s2t*sl];
41 end
42
43 function y2 = PSAG(t,del)
44 c2t=cos(2*t);
45 s2t=sin(2*t);
46 sl=sin(del);
47 cl=cos(del);
48 y2=[1,-((c2t^2)+((s2t^2)*cl)),-c2t*s2t*(1-cl),s2t*sl];
49 end

26
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

For further resolving the angles (more accurate results):

1 function resolver()
2 del=pi/2;
3 i=0;
4 data=load('DetW.txt');
5 [p,q]=size(data);
6 for j=1:p
7 disp ([num2str(j),' of ',num2str(p)]);
8 for t1=(data(j,1)-1):0.2:(data(j,1)+1)
9 for t2=(data(j,2)-1):0.2:(data(j,2)+1)
10 for t3=(data(j,3)-1):0.2:(data(j,3)+1)
11 for t4=(data(j,4)-1):0.2:(data(j,4)+1)
12 PSG(:,1)=PSGG(t1/180*pi,del);
13 PSG(:,2)=PSGG(t2/180*pi,del);
14 PSG(:,3)=PSGG(t3/180*pi,del);
15 PSG(:,4)=PSGG(t4/180*pi,del);
16 PSA(1,:)=PSAG(t1/180*pi,del);
17 PSA(2,:)=PSAG(t2/180*pi,del);
18 PSA(3,:)=PSAG(t3/180*pi,del);
19 PSA(4,:)=PSAG(t4/180*pi,del);
20 W=kron(PSA,PSG');
21 DW=det(W);
22 i=i+1;
23 detW(i,:)=[t1,t2,t3,t4,DW];
24 clear W
25 End
26 End
27 End
28 End
29 [M,in]=max(detW(:,5));
30 detW(in,:)
31 clear detW
32 End
33 End
34
35 function y1 = PSGG(t,del)
36 c2t=cos(2*t);
37 s2t=sin(2*t);
38 sl=sin(del);
39 cl=cos(del);
40 y1=[1;(c2t^2)+((s2t^2)*cl);c2t*s2t*(1-cl);s2t*sl];
41 End
42
43 function y2 = PSAG(t,del)
44 c2t=cos(2*t);
45 s2t=sin(2*t);
46 sl=sin(del);
47 cl=cos(del);
48 y2=[1,-((c2t^2)+((s2t^2)*cl)),-c2t*s2t*(1-cl),s2t*sl];
49 End

27
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Code 2:

The following code is for calculating the Mueller matrix for the sample (with blank correction) whose data
was recorded in the form of images for a particular wavelength of incident light. The algorithm is as
described briefly above in the Analysis section of this report.

1 clear; clc;
2 cd Blank;
3 hh=sum(sum(imread('hh.jpg')));
4 hp=sum(sum(imread('hp.jpg')));
5 hv=sum(sum(imread('hr.jpg')));
6 hr=sum(sum(imread('hv.jpg')));
7 ph=sum(sum(imread('ph.jpg')));
8 pp=sum(sum(imread('pp.jpg')));
9 pr=sum(sum(imread('pr.jpg')));
10 pv=sum(sum(imread('pv.jpg')));
11 rh=sum(sum(imread('rh.jpg')));
12 rp=sum(sum(imread('rp.jpg')));
13 rr=sum(sum(imread('rr.jpg')));
14 rv=sum(sum(imread('rv.jpg')));
15 vh=sum(sum(imread('vh.jpg')));
16 vp=sum(sum(imread('vp.jpg')));
17 vr=sum(sum(imread('vr.jpg')));
18 vv=sum(sum(imread('vv.jpg')));
19 cd ..;
20
21 PSG=[1,1,1,1;
22 1,-1,0,0;
23 0,0,1,0;
24 0,0,0,1];
25 PSA=PSG';
26 W=kron(PSA,PSG');
27 Winv=inv(W);
28
29 Mi_b=[hh;hv;hp;hr;vh;vv;vp;vr;ph;pv;pp;pr;rh;rv;rp;rr];
30 Ms_b=Winv*Mi_b;
31 Ms_b=Ms_b/Ms_b(1,1);
32 Ms_b=reshape(Ms_b,4,4)'
33
34 cd QWP_45;
35 hh=sum(sum(imread('hh.jpg')));
36 hp=sum(sum(imread('hp.jpg')));
37 hv=sum(sum(imread('hr.jpg')));
38 hr=sum(sum(imread('hv.jpg')));
39 ph=sum(sum(imread('ph.jpg')));
40 pp=sum(sum(imread('pp.jpg')));
41 pr=sum(sum(imread('pr.jpg')));
42 pv=sum(sum(imread('pv.jpg')));
43 rh=sum(sum(imread('rh.jpg')));
44 rp=sum(sum(imread('rp.jpg')));
45 rr=sum(sum(imread('rr.jpg')));
46 rv=sum(sum(imread('rv.jpg')));
47 vh=sum(sum(imread('vh.jpg')));
48 vp=sum(sum(imread('vp.jpg')));
49 vr=sum(sum(imread('vr.jpg')));
50 vv=sum(sum(imread('vv.jpg')));
51 cd ..;
52
53 Mi_QWP=[hh;hv;hp;hr;vh;vv;vp;vr;ph;pv;pp;pr;rh;rv;rp;rr];

28
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

54 Ms_QWP=Winv*Mi_QWP;
55 Ms_QWP=Ms_QWP/Ms_QWP(1,1);
56 Ms_QWP=reshape(Ms_QWP,4,4)';
57 Ms_QWP=Ms_QWP*inv(Ms_b);
58 Ms_QWP=Ms_QWP/Ms_QWP(1,1)

Code 3:

The following code was used to analyze the data obtained in the form of spectra for white light. It
calculates the Mueller matrix and then uses a decomposition scheme to decompose it into depolarization,
di-attenuation and linear retardance for each wavelength of light.

1 clear; clc;
2 dummy=load('3535.txt');
3 mvec(1,:,:)=load('3535.txt');
4 mvec(2,:,:)=load('3570.txt');
5 mvec(3,:,:)=load('35105.txt');
6 mvec(4,:,:)=load('35140.txt');
7 mvec(5,:,:)=load('7035.txt');
8 mvec(6,:,:)=load('7070.txt');
9 mvec(7,:,:)=load('70105.txt');
10 mvec(8,:,:)=load('70140.txt');
11 mvec(9,:,:)=load('10535.txt');
12 mvec(10,:,:)=load('10570.txt');
13 mvec(11,:,:)=load('105105.txt');
14 mvec(12,:,:)=load('105140.txt');
15 mvec(13,:,:)=load('14035.txt');
16 mvec(14,:,:)=load('14070.txt');
17 mvec(15,:,:)=load('140105.txt');
18 mvec(16,:,:)=load('140140.txt');
19
20 mvecb(1,:,:)=load('b3535.txt');
21 mvecb(2,:,:)=load('b3570.txt');
22 mvecb(3,:,:)=load('b35105.txt');
23 mvecb(4,:,:)=load('b35140.txt');
24 mvecb(5,:,:)=load('b7035.txt');
25 mvecb(6,:,:)=load('b7070.txt');
26 mvecb(7,:,:)=load('b70105.txt');
27 mvecb(8,:,:)=load('b70140.txt');
28 mvecb(9,:,:)=load('b10535.txt');
29 mvecb(10,:,:)=load('b10570.txt');
30 mvecb(11,:,:)=load('b105105.txt');
31 mvecb(12,:,:)=load('b105140.txt');
32 mvecb(13,:,:)=load('b14035.txt');
33 mvecb(14,:,:)=load('b14070.txt');
34 mvecb(15,:,:)=load('b140105.txt');
35 mvecb(16,:,:)=load('b140140.txt');
36
37 wavel = dummy(:,1);
38 lim = find(mvec(1,:,1)>=470 & mvec(1,:,1)<=840);
39 a = lim(1,1);
40 b = lim(1,end);
41 for m2 = a:b,
42 muela=mueller_psa(mvec(:,m2,2)');
43 muelb=mueller_psa(mvecb(:,m2,2)');
44 muela = muela / muela(1,1);
45 muelb = muelb / muelb(1,1);

29
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

46 muel = muela*inv(muelb);
47 muel = muel/muel(1,1);
48 flu_mat(m2,:,:)=muel;
[rvec,dvec,R(m2),D(m2),depol(m2),rotation(m2),lin_reta(m2)] =
49
polardecomposition_paperm(muel);
50 end
51 totint=(flu_mat(a:b,1,1));
52 subplot(4,4,1)
53 plot(flu_mat(a:b,1,1))
54 subplot(4,4,2)
55 plot(flu_mat(a:b,1,2))
56 subplot(4,4,3)
57 plot(flu_mat(a:b,1,3))
58 subplot(4,4,4)
59 plot(flu_mat(a:b,1,4))
60 subplot(4,4,5)
61 plot(flu_mat(a:b,2,1))
62 subplot(4,4,6)
63 plot(flu_mat(a:b,2,2))
64 subplot(4,4,7)
65 plot(flu_mat(a:b,2,3))
66 subplot(4,4,8)
67 plot(flu_mat(a:b,2,4))
68 subplot(4,4,9)
69 plot(flu_mat(a:b,3,1))
70 subplot(4,4,10)
71 plot(flu_mat(a:b,3,2))
72 subplot(4,4,11)
73 plot(flu_mat(a:b,3,3))
74 subplot(4,4,12)
75 plot(flu_mat(a:b,3,4))
76 subplot(4,4,13)
77 plot(flu_mat(a:b,4,1))
78 subplot(4,4,14)
79 plot(flu_mat(a:b,4,2))
80 subplot(4,4,15)
81 plot(flu_mat(a:b,4,3))
82 subplot(4,4,16)
83 plot(flu_mat(a:b,4,4))
84 figure
85 plot(depol(a:b))
86 figure
87 plot(D(a:b))
88 figure
89 plot(lin_reta(a:b))
90 figure
91 plot(rotation(a:b))
92 depol = depol';
93 D = D';
94 lin_reta = lin_reta';
95 rotation = rotation';
96 wavel = wavel;
97 allpol = [wavel(a:b) depol(a:b) D(a:b) lin_reta(a:b) rotation(a:b)];
98 save 'polpara.dat' allpol -ascii;

30
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Included Functions:

 Mueller_psa()

1 function[Muel]=mueller_psa(m_vec)
2 psg = [1.0000 1.0000 1.0000 1.0000
3 0.1169 0.5870 0.7498 0.0302
4 0.3213 -0.4924 0.4331 -0.1713
5 0.9397 0.6427 -0.5002 -0.9848];
6 psa = [1.0000 -0.1169 -0.3213 0.9387
7 1.0000 -0.5870 0.4924 0.6427
8 1.0000 -0.7498 -0.4331 -0.5002
9 1.0000 -0.0302 0.1713 -0.9848];
10 m_t=kron(psa, psg');
11 m_trans=inv(m_t);
12 M=m_trans*m_vec';
13 Muel=[M(1:4)';M(5:8)';M(9:12)';M(13:16)']'/M(1,1);

 Polardecomposition_paperm()

1 function[rvec,dvec,R,D,depol,rotation,lin_reta]=polardecomposition_paperm(muel)
2 format long
3 I=[1 0 0;
4 0 1 0;
5 0 0 1];
6 pvec=[muel(2,1),muel(3,1),muel(4,1)]*(1/muel(1,1));
7 dvec=[muel(1,2),muel(1,3),muel(1,4)]*(1/muel(1,1));
8 D=((muel(1,2)^2+muel(1,3)^2+muel(1,4)^2)^0.5)*(1/muel(1,1));
9 m=(1/muel(1,1))*[muel(2,2),muel(2,3),muel(2,4);
10 muel(3,2),muel(3,3),muel(3,4);
11 muel(4,2),muel(4,3),muel(4,4)];
12 D1=(1-D^2)^0.5;
13 if D==0
14 muel_0=muel/muel(1,1);
15 else
16 mD=D1*I+(1-D1)*dvec'*dvec/D^2;
17 MD=muel(1,1)*[1,dvec;
18 dvec',mD]
19 diattenuation = ((MD(1,2)^2+MD(1,3)^2+MD(1,4)^2)^0.5)*(1/MD(1,1))
20 muel_0=muel*inv(MD);
21 end
22 m_1=[muel_0(2,2) muel_0(2,3) muel_0(2,4);
23 muel_0(3,2) muel_0(3,3) muel_0(3,4);
24 muel_0(4,2) muel_0(4,3) muel_0(4,4)];
25 l_0=eig(m_1*m_1');
26 m_0=inv(m_1*m_1'+((l_0(1)*l_0(2))^0.5+(l_0(2)*l_0(3))^0.5+(l_0(3)*l_0(1))^0.5)*I);
27 m_00=(l_0(1)^0.5+l_0(2)^0.5+l_0(3)^0.5)*m_1*m_1'+I*(l_0(1)*l_0(2)*l_0(3))^0.5;
28 if det(m_1)>=0
29 mdelta=m_0*m_00;
30 else
31 mdelta=-m_0*m_00;
32 end
33 [v,mdeltaf] = eig(mdelta);
34 depol=1-(abs(mdelta(1,1))+abs(mdelta(2,2))+abs(mdelta(3,3)))/3
35 depol1 =1-(abs(mdeltaf(1,1))+abs(mdeltaf(2,2))+abs(mdeltaf(3,3)))/3
36 nul=(pvec'-m*dvec')/D1^2;
37 Mdelta=[1 0 0 0;

31
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

38 nul mdelta]
39 Mdeltaf =[1 0 0 0;
40 nul mdeltaf]
41 Mdinv=inv(Mdelta);
42 MR=Mdinv*muel_0
43 trmR=(MR(2,2)+MR(3,3)+MR(4,4))/2;
44 argu=trmR-1/2;
45 if abs(argu)>1
46 if argu>0
47 R=acos(1);
48 else
49 R=acos(-1);
50 end
51 else
52 R=acos(argu);
53 end
54 cssq_10=(MR(2,2)+MR(3,3))^2+(MR(3,2)-MR(2,3))^2;
55 tan_rot=(MR(3,2)-MR(2,3))/(abs(MR(2,2))+abs(MR(3,3)));
56 de=cssq_10^0.5-1;
57 if de>0.999999999999
58 de=1;
59 end
60 if de<-0.99999999999
61 de=-1;
62 end
63 lin_reta=acos(de);
64 rotation=0.5*atan(tan_rot);
65 if tan_rot<-0.000000001
66 rotation=rotation+pi;
67 end
68 if abs(MR(3,2)-MR(2,3))<=0.000000001 & abs(MR(2,2)+MR(3,3))>0.0000000001
69 rotation=0;
70 end
71 if abs(sin(R))<=0.000000001
72 a3=((1+cos(lin_reta))/2)^0.5;
73 a1=(MR(3,4)+MR(4,3))/(4*a3);
74 a2=(MR(4,2)+MR(2,4))/(4*a3);
75 else
76 D2=1/(2*sin(R));
77 a1=D2*(MR(3,4)-MR(4,3));
78 a2=D2*(MR(4,2)-MR(2,4));
79 a3=D2*(MR(2,3)-MR(3,2));
80 end
81 rvec=[1,a1,a2,a3]'
82 if abs(cos(R))>=0.9999999999
83 C1=MR(2,2)+MR(3,3);
84 C2=MR(2,3)-MR(3,2);
85 if abs(C1)<0.0000000001
86 MR=MR*[1 0 0 0; 0 1 0 0; 0 0 -1 0; 0 0 0 -1];
87 lin_reta=pi;
88 end
89 end
90 orientation = 0.5*acos(MR(3,4)/sin(lin_reta))
91 return

32
Indian Institute of Science Education and Research, Kolkata 28th March – 25th April 2011

Notes / Remarks:

33