• Tidak ada hasil yang ditemukan

Chapter VI: Diffusing wave spectroscopy: a unified treatment on temporal

6.6 Appendix

SNR of decorrelation time measurements in spatial ensemble methods

When the light is reflected from the dynamic sample, the light intensity at position π‘Ÿ and time𝑑, πΌπ‘Ÿ(𝑑), can be decoupled into two parts,

πΌπ‘Ÿ(𝑑) =πΌπ‘Ÿ ,𝑆(𝑑) +𝑛(𝑑), (6.3) where πΌπ‘Ÿ ,𝑆(𝑑) is the intensity of one speckle of the signal light that is perturbed by the scattering media, and𝑛(𝑑) is the intensity fluctuation from noise, such as shot

noise and detector noise. By the definition of noise, 𝑛(𝑑) has zero mean. πΌπ‘Ÿ ,𝑆(𝑑) follows exponential distribution due to speckle statistics [17]. For convenience, we define the AC part ofπΌπ‘Ÿ ,𝑆(𝑑)andπΌπ‘Ÿ(𝑑)as ΛœπΌπ‘Ÿ ,𝑆(𝑑)and ΛœπΌπ‘Ÿ(𝑑), respectively, therefore we have

˜

πΌπ‘Ÿ(𝑑) =πΌΛœπ‘Ÿ ,𝑆(𝑑) +𝑛(𝑑). (6.4) Here, both ΛœπΌπ‘Ÿ ,𝑆(𝑑)and ΛœπΌπ‘Ÿ(𝑑)are zero mean, and

hπΌΛœπ‘Ÿ ,𝑆(𝑑)i = q

hπΌΛœπ‘Ÿ ,𝑆(𝑑)i = 𝐼0 (6.5) due to speckle statistics [17]. Here, hΒ·i denotes the expected value and 𝐼0 is the expected value of πΌπ‘Ÿ ,𝑆(𝑑).

Define the signalπ‘†π‘Ÿ recorded on the camera pixel located at positionπ‘Ÿ as π‘†π‘Ÿ =

∫ 𝑇 0

𝛼 πΌπ‘Ÿ(𝑑)𝑑 𝑑 , (6.6)

where𝛼is the factor that relates the photon numbers to photoelectrons on camera pixels, including detector quantum efficiency, light collection efficiency, and other experimental imperfections, and𝑇 is the camera exposure time.

The speckle contrast𝛾among the whole camera frame is defined as 𝛾 =

s

h(π‘†π‘Ÿβˆ’ hπ‘†π‘Ÿi)2i

hπ‘†π‘Ÿi2 or𝛾2= h(π‘†π‘Ÿ βˆ’ hπ‘†π‘Ÿi)2i hπ‘†π‘Ÿi2

. (6.7)

The numerator of the𝛾2is 𝛾2

𝑒 𝑝 =h(π‘†π‘Ÿ βˆ’ hπ‘†π‘Ÿi)2i

=h(

∫ 𝑇

0

π›ΌπΌΛœπ‘Ÿ(𝑑)𝑑 𝑑)2i

=h

∫ 𝑇

0

∫ 𝑇

0

𝛼2πΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)𝑑 𝑑1𝑑 𝑑2i.

(6.8)

Substituting equation 6.4 into Eq. 6.8, we have

𝛾2

𝑒 𝑝 =h

∫ 𝑇

0

∫ 𝑇

0

𝛼2πΌΛœπ‘Ÿ ,𝑆(𝑑1)πΌΛœπ‘Ÿ ,𝑆(𝑑2)𝑑 𝑑1𝑑 𝑑2i + h

∫ 𝑇

0

∫ 𝑇

0

𝛼2𝑛(𝑑1)𝑛(𝑑2)𝑑 𝑑1𝑑 𝑑2i

=𝛼2h𝐼˜2

π‘Ÿi

∫ 𝑇

0

∫ 𝑇

0

𝑔𝑆(𝑑1βˆ’π‘‘2)𝑑 𝑑1𝑑 𝑑2+𝛼2h𝑛2i

∫ 𝑇

0

∫ 𝑇

0

𝑔𝑛(𝑑1βˆ’π‘‘2)𝑑 𝑑1𝑑 𝑑2

=𝛼2h𝐼˜2

π‘Ÿi𝑇

∫ 𝑇

0

2(1βˆ’ 𝑑 𝑇

)𝑔𝑆(𝑑)𝑑 𝑑+𝛼2h𝑛2i𝑇

∫ 𝑇

0

2(1βˆ’ 𝑑 𝑇

)𝑔𝑛(𝑑)𝑑 𝑑 .

(6.9)

Here, 𝑔𝑆(𝑑) is the correlation function of the mean-removed signal light intensity, and 𝑔𝑛(𝑑) is the correlation function of noise. If we assume 𝑔𝑆(𝑑) = π‘’βˆ’2𝑑/𝜏 and 𝑔𝑛(𝑑) = π‘’βˆ’π‘‘/πœπ‘›, where𝜏is the decorrelation time of the speckle decorrelation time and πœπ‘› (related to the detector bandwidth BW, 1/BW) is the noise decorrelation time, the above equation can be simplified as

𝛾2

𝑒 𝑝 =𝛼2h𝐼˜2

π‘Ÿi𝑇 𝜏+2𝛼2h𝑛2i𝑇 πœπ‘›. (6.10) If the detector is working under the shot noise dominant scheme, where the mean of the number of photoelectrons is equal to the standard deviation of the number of photoelectrons, we have

𝛼 𝐼0𝑇 =h(

∫ 𝑇

0

𝛼𝑛(𝑑)𝑑 𝑑)2i

=2𝛼2h𝑛2i𝑇 πœπ‘›.

(6.11)

Substituting the above equation and Eq. 6.5 into Eq. 6.10, the numerator of the contrast square can be further simplified as

𝛾2

𝑒 𝑝=𝛼2𝐼2

0𝑇 𝜏+𝛼 𝐼0𝑇 . (6.12)

The denominator of𝛾 is

π›Ύπ‘‘π‘œπ‘€ 𝑛 =hπ‘†π‘Ÿi=𝛼 𝐼0𝑇 . (6.13)

Hence, the contrast has the expression of 𝛾2=

𝛾2

𝑒 𝑝

𝛾2

π‘‘π‘œπ‘€ 𝑛

= 𝛼2𝐼2

0𝑇 𝜏+𝛼 𝐼0𝑇 (𝛼 𝐼0𝑇)2

= 𝜏 𝑇

+ 1

𝛼 𝐼0𝑇

= 𝜏 𝑇

+ 1 𝑁𝑇

(6.14)

where𝑁𝑇 is the number of the photoelectrons in one speckle within the camera ex- posure time. Conventional speckle statistics without considering shot noise predicts that the speckle contrast scales with respect to 1/p

𝑁𝑝 π‘Žπ‘‘ 𝑑 π‘’π‘Ÿ 𝑛, where 𝑁𝑝 π‘Žπ‘‘ 𝑑 π‘’π‘Ÿ 𝑛 is the number of independent decorrelation patterns recorded by the camera sensor within the exposure time. Intuitively,𝑁𝑝 π‘Žπ‘‘ 𝑑 π‘’π‘Ÿ 𝑛isβˆΌπ‘‡/𝜏since the ratio provides the number of decorrelation events within the camera exposure time. Here, the extra term 1/𝑁𝑇

in Eq. 6.14 is due to shot noise, i.e., depending on the photon budget. If the number of photoelectrons is sufficient, i.e., 1/𝑁𝑇 << 𝜏/𝑇 , we can discard this term and the expression degenerates to the conventional form.

In experiment, we can only collect a finite number of speckles and use the ensemble average to approximate the contrast. Hence, the contrast square calculated from one camera frame ˆ𝛾2is a statistical estimation:

Λ†

𝛾2 = h(π‘†π‘Ÿ βˆ’ hπ‘†π‘Ÿi)2i𝑓 𝑖𝑛𝑖𝑑 𝑒

hπ‘†π‘Ÿi𝑓 𝑖𝑛𝑖𝑑 𝑒 . (6.15)

Here, hΒ·i𝑓 𝑖𝑛𝑖𝑑 𝑒 denotes the ensemble average over the finite speckles in one camera frame. Therefore, both the numerator and denominator of the contrast square ˆ𝛾2are estimated from the finite speckles. To evaluate the accuracy of the estimation, we need to estimate the errors of both numerator and denominator in Eq. 6.15.

Given a random variable𝑋, if we use a sample average 1/𝑁𝑖𝑛𝑑𝑒 𝑝 𝑒𝑛𝑑𝑒𝑛𝑑

Í𝑁𝑖 𝑛𝑑 𝑒 𝑝 𝑒𝑛𝑑 𝑒𝑛𝑑

𝑖=1 𝑋𝑖

with𝑁𝑖𝑛𝑑𝑒 𝑝 𝑒𝑛𝑑𝑒𝑛𝑑 independent measurements to estimate its expected valueh𝑋i, the error between the sample average and the expected value is aboutp

𝑉(𝑋)/𝑁𝑖𝑛𝑑𝑒 𝑝 𝑒𝑛𝑑𝑒𝑛𝑑, where 𝑉(Β·) denotes the variance of the random variable 𝑋. In our calculation, 𝑁𝑖𝑛𝑑𝑒 𝑝 𝑒𝑛𝑑𝑒𝑛𝑑, the number of independent measurements (NIM) in spatial ensemble method, is the number of speckle grains in the camera frame, which is termed 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™.

Let us first calculate the variance of the numerator (𝛾2

𝑒 𝑝) of the𝛾2. The variance of 𝛾2

𝑒 𝑝is 𝑉(𝛾2

𝑒 𝑝)= h(π‘†π‘Ÿ βˆ’ hπ‘†π‘Ÿi)4i βˆ’ h(π‘†π‘Ÿ βˆ’ hπ‘†π‘Ÿi)2i2

=𝛼4

∫ 𝑇 0

∫ 𝑇 0

∫ 𝑇 0

∫ 𝑇 0

hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑3)πΌΛœπ‘Ÿ(𝑑4)i𝑑 𝑑1𝑑 𝑑2𝑑 𝑑3𝑑 𝑑4

βˆ’π›Ό2h

∫ 𝑇

0

∫ 𝑇

0

˜

πΌπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)𝑑 𝑑1𝑑 𝑑2i.

(6.16)

The first term in the above equation takes the expected value of four random variables multiplied together. If Λ†πΌπ‘Ÿis a Gaussian random variable, the bracket can be expanded as

hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑3)πΌΛœπ‘Ÿ(𝑑4)i =hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)i hπΌΛœπ‘Ÿ(𝑑3)πΌΛœπ‘Ÿ(𝑑4)i + hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑3)i hπΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑4)i + hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑4)i hπΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑3)i.

(6.17) Here, even though ΛœπΌπ‘Ÿ is not a Gaussian random variable, we still take the formula as an approximation, and this approximation actually holds with tolerable errors based on our experimental results. Equation 6.16 then becomes

𝑉(𝛾2

𝑒 𝑝) β‰ˆ 𝛼4

∫ 𝑇

0

∫ 𝑇

0

∫ 𝑇

0

∫ 𝑇

0

(hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)i hπΌΛœπ‘Ÿ(𝑑3)πΌΛœπ‘Ÿ(𝑑4)i

+ hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑3)i hπΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑4)i + hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑4)i hπΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑3)i)𝑑 𝑑1𝑑 𝑑2𝑑 𝑑3𝑑 𝑑4

βˆ’π›Ό4h

∫ 𝑇

0

∫ 𝑇

0

˜

πΌπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)𝑑 𝑑1𝑑 𝑑2i

2

=2𝛼4h

∫ 𝑇 0

∫ 𝑇 0

˜

πΌπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)𝑑 𝑑1𝑑 𝑑2i

2

=2(𝛾2

𝑒 𝑝)2.

(6.18)

Therefore, if there are 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™ independent speckles in spatial ensemble methods, the numerator of ˆ𝛾2

𝑒 𝑝has a form of Λ†

𝛾2

𝑒 𝑝 =𝛾2

𝑒 𝑝±

√ 2𝛾2

𝑒 𝑝

p𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

=𝛾2

𝑒 𝑝(1Β± s

2 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

). (6.19)

Here, the term after theΒ±denotes the standard error of the statistical estimation.

Next, let us calculate the error of the denominator (π›Ύπ‘‘π‘œπ‘€ 𝑛) of𝛾. It is simply s

𝑉(π‘†π‘Ÿ) 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

= s

h(π‘†π‘Ÿ βˆ’ hπ‘†π‘Ÿi)2i 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

= s

𝛾𝑒 𝑝2 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

. (6.20)

Therefore, the denominator ˆ𝛾2

π‘‘π‘œπ‘€ 𝑛has a form of Λ†

𝛾2

π‘‘π‘œπ‘€ 𝑛 = (π›Ύπ‘‘π‘œπ‘€ 𝑛± s

𝛾2

𝑒 𝑝

𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

)2 β‰ˆπ›Ύ2

π‘‘π‘œπ‘€ 𝑛± 2𝛾𝑒 π‘π›Ύπ‘‘π‘œπ‘€ 𝑛 p𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

. (6.21)

Finally, by combining equations 6.15, 6.19, and 6.21, the expression of the estimation of ˆ𝛾2is

Λ†

𝛾2= 𝛾ˆ2

𝑒 𝑝

Λ† 𝛾2

π‘‘π‘œπ‘€ 𝑛

β‰ˆ 𝛾2

𝑒 𝑝

𝛾2

π‘‘π‘œπ‘€ 𝑛

(1Β± s

1 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

vt

2+ 4𝛾2

𝑒 𝑝

𝛾2

π‘‘π‘œπ‘€ 𝑛

)

=(𝜏 𝑇

+ 1 𝑁𝑇

) (1Β± s

1 𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

r

2+4(𝜏 𝑇

+ 1 𝑁𝑇

)).

(6.22)

Hence, the estimation of the contrast is Λ†

𝛾 = r

𝜏 𝑇

+ 1 𝑁𝑇

(1Β± 1 2

s 1

𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™ r

2+4(𝜏 𝑇

+ 1 𝑁𝑇

)). (6.23)

In SVS, we usually set the camera exposure𝑇 much greater than the decorrelation time𝜏, e.g.,𝑇 >> 𝜏, and the number of photons collected by one camera pixel 𝑁𝑇

is also much greater than 1, e.g., 𝑁𝑇 >> 1. In this case, in the above equation, the second term in the second square root in the error part can be dropped and the estimation of the contrast square ˆ𝛾 can be approximated as

Λ† 𝛾 =

r 𝜏 𝑇

+ 1 𝑁𝑇

(1Β± s

1 2𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

). (6.24)

Rewriting the above equation, we have

𝜏 =𝑇𝛾ˆ2(1Β± s

1 2𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

) βˆ’ 𝑇 𝑁𝑇

. (6.25)

The SNR of the decorrelation time in spatial ensemble methods is 𝑆 𝑁 𝑅𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™ = 𝜏

π‘’π‘Ÿ π‘Ÿ(𝜏) = 𝜏 𝑇𝛾ˆ2

q 1

2𝑀𝑠 𝑝 π‘Žπ‘‘ 𝑖 π‘Žπ‘™

= 1 1+ 𝑇

𝜏 𝑁𝑇

r

𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™

2 .

(6.26)

Here, π‘’π‘Ÿ π‘Ÿ(𝜏) is the standard error of 𝜏, which is equal to𝑇𝛾ˆ2

q 1

2𝑀𝑠 𝑝 π‘Žπ‘‘ 𝑖 π‘Žπ‘™. As we defineπ‘πœ as the number of photoelectrons on each camera pixel per time interval𝜏, we findπ‘πœ = 𝑁𝑇

𝑇

𝜏. The above Eq. 6.26 can be simplified as 𝑆 𝑁 𝑅𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™ = 1

√ 2

1 1+ 1

π‘πœ

p

𝑀𝑠 𝑝 π‘Žπ‘‘π‘– π‘Žπ‘™. (6.27)

SNR of decorrelation time measurements in temporal sampling methods In temporal sampling methods, a fast photodetector with a sufficient bandwidth, such as a single-photon-counting-module (SPCM), is used to well sample the temporal traceπΌπ‘Ÿ(𝑑), and the decorrelation time𝜏is computed from the intensity correlation function𝐺2(𝑑):

𝐺2(𝑑)= 1 𝑇

𝛼2

∫ 𝑇 0

πΌπ‘Ÿ(𝑑1)πΌπ‘Ÿ(𝑑1βˆ’π‘‘)𝑑 𝑑1. (6.28) In practice, the correlation is performed between the mean-removed intensity fluc- tuation:

𝐺˜2(𝑑) = 1 𝑇

𝛼2

∫ 𝑇

0

˜

πΌπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑1βˆ’π‘‘)𝑑 𝑑1, (6.29) where ˜𝐺2(𝑑) denotes the intensity correlation function of the two mean-removed intensity traces, ΛœπΌπ‘Ÿ(𝑑)is the AC part of the intensity fluctuation,𝑑1is the time variable for integral, and𝑑is the time offset between the two intensity traces.

By substituting Eq. 6.4 into Eq. 6.29, we have

˜

𝐺2(𝑑) = 1 𝑇

𝛼2

∫ 𝑇

0

[πΌΛœπ‘Ÿ ,𝑆(𝑑1) +𝑛(𝑑1)] [πΌΛœπ‘Ÿ ,𝑆(𝑑1βˆ’π‘‘) +𝑛(𝑑1βˆ’π‘‘)]𝑑 𝑑1. (6.30) The expected value of ˜𝐺2(𝑑) is

h𝐺˜2(𝑑)i =𝛼2𝐼2

0𝑔𝑆(𝑑) +𝛼2h𝑛2i𝑔𝑛(𝑑). (6.31) Same as the definition before,𝑔𝑆(𝑑)is the correlation function of the mean-removed signal light intensity, and𝑔𝑛(𝑑)is the correlation function of noise.

When we use the finite time average to estimate the expected value of ˜𝐺2(𝑑), we need to calculate the variance of ˜𝐺2(𝑑):

𝑉[𝐺˜2(𝑑)] = 1 𝑇

(h𝐺˜2(𝑑)2i βˆ’ h𝐺˜2(𝑑)i2)

= 1 𝑇

h

∫ 𝑇

0

∫ 𝑇

0

𝛼4πΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑1βˆ’π‘‘)πΌΛœπ‘Ÿ(𝑑2)πΌΛœπ‘Ÿ(𝑑2βˆ’π‘‘)𝑑 𝑑1𝑑 𝑑2i βˆ’ 1 𝑇2

h𝐺˜2(𝑑)2i2

β‰ˆ 2𝛼4 𝑇2

∫ 𝑇

0

∫ 𝑇

0

hπΌΛœπ‘Ÿ(𝑑1)πΌΛœπ‘Ÿ(𝑑2)i2𝑑 𝑑1𝑑 𝑑2

= 2𝛼4 𝑇2

∫ 𝑇

0

∫ 𝑇

0

[𝐼2

0𝑔𝑆(𝑑) + h𝑛2i𝑔𝑛(𝑑)]2𝑑 𝑑1𝑑 𝑑2

β‰ˆ2(𝛼4𝐼4

0

𝜏 2𝑇

+𝛼3𝐼3

0

1 𝑇

+𝛼2𝐼2

0

1 𝑇 πœπ‘›

).

(6.32)

Hence, if we calculate the correlation function ˜𝐺2(𝑑) by using a finite long mea- surement trace and use it to estimate h𝐺˜2(𝑑)i, we have the following estimation form

𝐺˜2(𝑑) =h𝐺˜2(𝑑)i Β± q

𝑉[𝐺˜2(𝑑)]

= [𝛼2𝐼2

0𝑔𝑆(𝑑) +𝛼2h𝑛2i𝑔𝑛(𝑑)] Β± r

2(𝛼4𝐼4

0

𝜏 2𝑇

+𝛼3𝐼3

0

1 𝑇

+𝛼2𝐼2

0

1 𝑇 πœπ‘›

).

(6.33)

Since 𝑔𝑛(𝑑) usually has much shorter decorrelation time compared to 𝑔𝑆(𝑑), to estimate the speckle decorrelation time 𝜏, we can use the part of the correlation curve where𝑔𝑛(𝑑)4 drops close to 0 while 𝑔𝑆(𝑑) is still close to unity. In this case,

the part of the the correlation curve ˆ𝐺2(𝑑)is 𝐺ˆ2(𝑑) = [𝛼2𝐼2

0𝑔𝑆(𝑑) Β± r

2(𝛼4𝐼4

0

𝜏 2𝑇

+𝛼3𝐼3

0

1 𝑇

+𝛼2𝐼2

0

1 𝑇 πœπ‘›

)

=𝛼2𝐼2

0[𝑔𝑆(𝑑) Β± s

2( 𝜏 2𝑇

+ 1

𝛼 𝐼0𝑇

+ 1

𝛼2𝐼2

0𝑇 πœπ‘› )].

(6.34)

In the experiment,πœπ‘›can be approximated as the inverse of the detector bandwidth, or equivalently the time interval Δ𝑇 between two data points. In the following calculation, we will substituteπœπ‘›byΔ𝑇.

When we use the decorrelation curve to estimate a parameter associated with the curve, such as decorrelation time, there exist different fitting models to retrieve the parameter. Here, for simplicity, the estimated decorrelation time Λ†πœcan be chosen by taking the time point where the decorrelation curve drops to 1/𝑒. In this case, the error of the estimated decorrelation timeπ‘’π‘Ÿ π‘Ÿ(𝜏)is

π‘’π‘Ÿ π‘Ÿ(𝜏) = 1

|𝑑𝑔𝑆

𝑑 𝑑 |𝑔𝑆(𝑑)=1/𝑒| s

2( 𝜏 2𝑇

+ 1

𝛼 𝐼0𝑇

+ 1

𝛼2𝐼2

0𝑇Δ𝑇 )

= 𝑒 2𝜏

s 2( 𝜏

2𝑇

+ 1

𝛼 𝐼0𝑇

+ 1

𝛼2𝐼2

0𝑇Δ𝑇 ).

(6.35)

Hence, the decorrelation time𝜏can be estimated from the calculated decorrelation time𝜏as

𝜏=πœΛ†(1Β± 𝑒

√ 2

s 𝜏 2𝑇

+ 1

𝛼 𝐼0𝑇

+ 1

𝛼2𝐼2

0𝑇Δ𝑇

). (6.36)

The SNR of the decorrelation time in temporal sampling methods is 𝑆 𝑁 𝑅𝑑 π‘’π‘š 𝑝 π‘œπ‘Ÿ π‘Žπ‘™ = 𝜏

π‘’π‘Ÿ π‘Ÿ(𝜏) =

√ 2 𝑒

1 q𝜏

2𝑇 + 1

𝛼 𝐼0𝑇 + 1

𝛼2𝐼2

0𝑇Δ𝑇

. (6.37)

As defined in the main text, the NIM in temporal domain methods 𝑀𝑑 π‘’π‘š 𝑝 π‘œπ‘Ÿ π‘Žπ‘™ = 2𝑇

𝜏 , and taking the fact that 𝛼 𝐼0𝑇 = 12𝑀𝑑 π‘’π‘š 𝑝 π‘œπ‘Ÿ π‘Žπ‘™π‘πœ, the SNR equation 6.37 can be rewritten as

𝑆 𝑁 𝑅𝑑 π‘’π‘š 𝑝 π‘œπ‘Ÿ π‘Žπ‘™ =

√ 2 𝑒

1 q1+ 2

𝑁𝑇 + 2

𝑁2

𝑇

𝜏 Δ𝑇

p

𝑀𝑑 π‘’π‘š 𝑝 π‘œπ‘Ÿ π‘Žπ‘™. (6.38)

References

[1] D. J. Pine, D. A. Weitz, P. M. Chaikin, and E. Herbolzheimer. Diffusing wave spectroscopy.Physical Review Letters, 60(12):1134–1137, March 1988.

[2] JΓΆrg Stetefeld, Sean A. McKenna, and Trushar R. Patel. Dynamic light scat- tering: a practical guide and applications in biomedical sciences.Biophysical Reviews, 8(4):409–427, October 2016.

[3] Theodore J. Huppert, Solomon G. Diamond, Maria A. Franceschini, and David A. Boas. HomER: a review of time-series analysis methods for near- infrared spectroscopy of the brain.Applied Optics, 48(10):D280, March 2009.

[4] Gerard M. Ancellet and Robert T. Menzies. Atmospheric correlation-time measurements and effects on coherent doppler lidar.Journal of the Optical Society of America A, 4(2):367, February 1987.

[5] D. A. Boas, L. E. Campbell, and A. G. Yodh. Scattering and imaging with diffusing temporal field correlations. Physical Review Letters, 75(9):1855–

1858, August 1995.

[6] Turgut Durduran and Arjun G. Yodh. Diffuse correlation spectroscopy for non-invasive, micro-vascular cerebral blood flow measurement.NeuroImage, 85:51–63, January 2014.

[7] T. Durduran, R. Choe, W. B. Baker, and A. G. Yodh. Diffuse optics for tissue monitoring and tomography.Reports on Progress in Physics, 73(7):076701, June 2010.

[8] Shuai Yuan, Anna Devor, David A. Boas, and Andrew K. Dunn. Determina- tion of optimal exposure time for imaging of blood flow changes with laser speckle contrast imaging.Applied Optics, 44(10):1823, April 2005.

[9] Mingjun Zhao, Chong Huang, Daniel Irwin, Siavash Mazdeyasna, Ahmed Bahrani, Nneamaka Agochukwu, Lesley Wong, and Guoqiang Yu. EMCCD- based speckle contrast diffuse correlation tomography of tissue blood flow distribution. In Biophotonics Congress: Biomedical Optics Congress 2018 (Microscopy/Translational/Brain/OTS). OSA, 2018.

[10] Andrew K. Dunn, Hayrunnisa Bolay, Michael A. Moskowitz, and David A.

Boas. Dynamic imaging of cerebral blood flow using laser speckle. Journal of Cerebral Blood Flow & Metabolism, 21(3):195–201, March 2001.

[11] David A. Boas and Andrew K. Dunn. Laser speckle contrast imaging in biomedical optics.Journal of Biomedical Optics, 15(1):011109, 2010.

[12] A. J. F. Siegert. On the fluctuations in signals returned by many indepen- dently moving scatterers, Radiation Laboratory, Massachusetts Insitute of Technology, 1943.

[13] R. Bandyopadhyay, A. S. Gittings, S. S. Suh, P. K. Dixon, and D. J. Durian.

Speckle-visibility spectroscopy: a tool to study time-varying dynamics. Re- view of Scientific Instruments, 76(9):093110, September 2005.

[14] Ichiro Inoue, Yuya Shinohara, Akira Watanabe, and Yoshiyuki Amemiya.

Effect of shot noise on x-ray speckle visibility spectroscopy.Optics Express, 20(24):26878, November 2012.

[15] Andrew K. Dunn. Laser speckle contrast imaging of cerebral blood flow.

Annals of Biomedical Engineering, 40(2):367–377, November 2011.

[16] Robert Zwanzig and Narinder K. Ailawadi. Statistical error due to finite time averaging in computer experiments.Physical Review, 182(1):280–283, June 1969.

[17] Joseph W. Goodman. Speckle Phenomena in Optics. Viva Books Private Limited, 2008.

[18] Detian Wang, Ashwin B. Parthasarathy, Wesley B. Baker, Kimberly Gannon, Venki Kavuri, Tiffany Ko, Steven Schenkel, Zhe Li, Zeren Li, Michael T.

Mullen, John A. Detre, and Arjun G. Yodh. Fast blood flow monitoring in deep tissues with real-time software correlators.Biomedical Optics Express, 7(3):776, February 2016.

[19] Wenjun Zhou, Oybek Kholiqov, Shau Poh Chong, and Vivek J. Srinivasan.

Highly parallel, interferometric diffusing wave spectroscopy for monitoring cerebral blood flow dynamics.Optica, 5(5):518, April 2018.

[20] J. Xu, A. K. Jahromi, J. Brake, J. E. Robinson, and C. Yang. Interferometric speckle visibility spectroscopy (isvs) for human cerebral blood flow monitor- ing, 2020. eprint:arXiv:2009.00002.