5th-6th Thermal and Fluids Engineering Conference (TFEC)

May 26–28, 2021

Virtual Conference

TFEC-2020-36282

Analytical Study of Phase Change Heat Transfer in Biological Tissue using

Coordinate Transformation Technique during Cryosurgery

Abhigyan Majumdar1, Sumit Kumar 1*

1
Department of Mechanical Engineering, National Institute of Technology, Rourkela-769008, Odisha, India

ABSTRACT

This study aims to develop the analytical solution to the phase change heat transfer process in biological tissue
undergoing cryosurgery. Cryosurgery is a clinical procedure for treating cancer by exposing the tumor tissues
to cryogenic temperatures. Studying the heat transfer process in the living tissue during cryosurgery can
provide insights on how to optimize the several parameters of the process and how to judge the effectiveness
of the process. Tracking the thermal history of the tumor tissue and healthy tissue and monitoring the freezing
front propagation can help maximize damage to cancer cells while minimizing damage to surrounding healthy
tissue. Involvement of phase change in the heat transfer process means the formation of a phase change
interface. Previous studies have used numerical schemes that are computationally intensive since the
temperature field, and the interface position are linked by non-linear parital differential equations. This study
proposes a novel approach using the coordinate transformation technique to reduce the non-linear partial
differential equations into ordinary differential equations. Firstly, a solution to the 1-D Stefan problem is
obtained and verified with previous results. Next, the Pennes bio-heat transfer equation is solved with the
isothermal phase change, and the results are presented. The results agree satisfactorily with previously
published work, and the computational time and resources required are negligible.

KEYWORDS: Bio-heat transfer, Cryosurgery, Phase change, Coordinate transformation technique, Analytical
solution, biological tissue

1. INTRODUCTION

The study of the phase change heat transfer process in biological tissue has diverse applications in the domains
of cryosurgery, cryopreservation, hypothermia, etc. [1, 2]. Cryosurgery is a clinical treatment procedure for
different types of tumors (benign and malignant), wherein the tumor tissue is killed by cooling it with the help
of a cryoprobe. This treatment methodology for cancer has gained greater attention in recent years, owing to
several advantages over others. Being a single application treatment procedure, cryosurgery does not allow
tumor tissue to progressively gain greater immunity against the treatment, which is often the case for other
treatment methods like chemotherapy, radiation, or hormonal deprivation therapy [3]. However, an accurate
understanding of the relation between various thermal characteristics and the clinical outcome of the surgeries
is needed for the success of cryosurgery as a treatment modality. During cryosurgery, the heat from the cancer
cells and adjoining healthy body tissue flows towards the cryoprobe, and the cancer cells thereby undergo
freezing and sub-cooling. The thermal history of the tumor tissues as well as the adjacent healthy tissues are
paramount to ascertain the level of damage done to the respective tissues. Since a real-time sensor-based
approach during the cryosurgery not only complicates the procedure but also has the potential to alter the
results of the surgery, a simulation-based approach to accurately predict the temperature distribution in and
the thermal history of the tumor and adjoining healthy tissues needs to be employed. A simulation-based
approach can also help us gain a better understanding of the dosage requirements of cryosurgery, i.e., how
much surgery time and what cryofluid temperature can produce exposure to low temperatures for enough time

*Corresponding Author: sumit1986.jaiswal@gmail.com

1
 TFEC-2020-36282

for the outcome to be most catastrophic for the cancer cells while doing least harm to healthy tissue around the
tumor. Simulation can also help optimize the relative location and orientations of cryoprobes in surgeries where
multiple cryoprobes are used to treat large and irregular shaped tumors. This has been studied in detail in [4,
5].

In past studies of the phase change heat transfer process during cryosurgery, researchers have employed many
numerical schemes to be able to accurately predict the freezing front propagation and temperature distribution
across the tissue. While some approaches call for interface tracking and subsequent solution of the energy
equations, other methods bypass the interface problem by enthalpy formulation or using the effective heat
capacity method. Fortin and Belhamadla [4] employed an anisotropic remeshing strategy to improve the
accuracy in a direct numerical solution of the phase change heat transfer process. The phase change interface
location was determined by a semi-phase field function, and the abrupt variation leads to oscillations in the
solution. To avoid this, the refinement of the mesh near to the interface needs to be increase. Hence after each
iteration, the mesh is refined, and the previous results are interpolated on the new mesh, thereby preserving
the accuracy of the solution. Other papers have solved the more complex problem involving non-ideal heat
transfer by considering the Hyperbolic heat transfer model or the Dual-Phase Lag Model. Deng and Liu [6]
attempted to solve the phase change heat transfer problem in combined cryosurgery and hyperthermia
treatment by direct numerical solution of the modified Pennes bioheat equation using the effective heat
capacity method. This removes the complexity of front tracking in a problem with multiple interfaces. Kumar
et al. [7] solved the Catteneo-Vernotte’s equation (hyperbolic bio-heat equation) in a 2D tissue domain with
different thermo-physical properties of healthy and cancer tissue regions, using the Enthalpy formulation
method. Not only does this bypass the need for interface tracking, but it also allowed the easy formulation of
the mushy zone between the 2 phases. The correlation between the relaxation time and the ice-ball propagation
rate was clearly understood from the study results. Singh and Kumar [8] studied the 1-D heat transfer in triple-
layer skin tissue using the Dual-Phase Lag Bio-heat model. The system of equations involving both the
relaxation and the thermalization phase lag constants was solved using the Enthalpy method, and the Dual-
Phase Lag Model was established as giving results intermediate of the parabolic and hyperbolic models.
Mochnacki and Majchrzak [9] solved the Dual-Phase Lag model using the effective heat capacity method in a
cylindrical domain. Ahmadikia and Moradi [10] have used the hyperbolic heat transfer model to study the
freezing phenomenon in biological tissue. The same authors have also used the Dual-Phase Lag Model to solve
the phase change heat transfer in biological tissue involving metabolic heat generation and blood perfusion
terms, using the enthalpy method.

It is found from the literature review that the researchers generally solved the phase change problem using the
numerical technique due to the non-linear partial differential equation. Therefore, the objective of the present
study is to develop an analytical solution for freezing/thawing problems. To the author’s best knowledge, it is
the first attempt to solve the solidification problem using the coordinate transformation technique. This study
proposes a novel approach using the coordinate transformation technique to reduce the non-linear partial
differential equations into ordinary differential equations. Therefore, it reduces the mathematical complexity
of the problem. Furthermore, it is computationally efficient than any numerical technique for phase change
problems. The analytical solution of the Pennes bio-heat transfer equation is simplified for the 1-D Stefan
problem. The results obtained using the analytical solution for the 1-D Stefan problem is verified with previous
results. After that, the Pennes bio-heat transfer equation is solved with the isothermal phase change, and the
results are presented. The results agree satisfactorily with previously published work, and the computational
time and resources required are negligible.

2. MATHEMATICAL MODEL

The generalized analytical solution to the phase change heat transfer process in biological tissue during
cryosurgery has been developed. After that, the analytical solution has been simplified for the Stefan
solidification problem in the 1-D semi-infinite domain to verify the novel technique developed in the present
study.

2
 TFEC-2020-36282

2.1 Mathematical Formulation of the phase change problem

Phase change heat transfer assumes a more complex form in the biological tissue because of the presence of
blood perfusion and metabolic heat generation (Eq. 1). The Pennes bio-heat transfer equation for 1-D
biological tissue is written as:
 2T T
k  bCb wb Tb  T   Q m  C (1)
x 2
t
The metabolic heat generation (the third term on the left-hand side of Eq. 1) is due to the biological activity of
the cells in the living tissue, while the blood perfusion terms (second term on the left-hand side of Eq. 1)
signifies the heat transferred from the blood to the cooler tissue region when it is being cooled by application
of cryoprobe. Thus, it is to be noted that both the terms are dependent on whether the tissue is biologically
active or not. It is widely accepted that both the heat generation terms cease to exist as soon as the tissue goes
into the frozen or mushy region [6, 7]. The tissue domain has been divided into two parts: the frozen and
unfrozen domain. It is assumed that the phase change occurs at a constant temperature. Therefore, the mushy
region has not been considered in the present study. Thus, Eq. (2) is followed in the frozen domain, while in
the unfrozen domain, Eq. (3) is used. The problem in the biological tissue can thus be defined as:

In frozen/solid domain:
 2T f ( x, t ) T f ( x, t )
kf   f Cf in 0 ≤ 𝑥 < 𝑠(𝑡); 𝑡 > 0 (2)
x 2 t
In unfrozen/liquid domain:
 2Tu ( x, t ) T ( x, t )
ku   b Cb wb Tb  Tu ( x, t )  Q m   u Cu u in 𝑠(𝑡) ≤ 𝑥 < ∞; 𝑡 > 0 (3)
x 2
t
The following boundary conditions (Eq. 4) and initial condition (Eq. 5) have been applied to find the unique
solutions of the above differential equations (Eqs. (2) and (3)).

T f ( x  0, t )  T0 (4a)
Tu ( x  , t )  Ti (4b)
Tu ( x  s (t ), t )  T f ( x  s (t ), t )  Tm (4c)
T f Tu ds (t )
kf  ku  f L (4d)
x x  s (t )
x x  s (t ) dt
Tu ( x, t  0)  Ti (5)

Equations (4c) and (4d) represent the continuity of temperature and heat flux at the frozen-unfrozen region
interface.

The solution of the Pennes bio-heat transfer equation for the freezing/thawing process was developed by
shifting the origin to the phase change interface. Since the interface itself moves with respect to time, any
variable defining the position with respect to the interface depends on both the position and the time variable.
This property is used to convert the 1-D partial differential equation into an ordinary differential equation with
respect to the new variable.

The new variable is defined as:

𝑦 (𝑥, 𝑡) = 𝑥 − 𝑠(𝑡) (6)

Hence, the new derivatives are obtained as,

3
 TFEC-2020-36282

T dT
 (7)
x dy
 2T d 2T
 (8)
x 2 dy 2
T dT ds(t )
 (9)
t dy dt

These relations are then used to convert the Eqs. (2) and (3) from a partial differential equation into an ordinary
differential equation in terms of T and y. The transformed governing equations is an ordinary differential
equation,
d 2T f ( y ) 1 ds(t ) dT f ( y)
 (10)
dy 2  f dt dy
d 2Tu ( y ) ds (t ) dTu ( y )
ku 2
  b C b wb Tb  Tu ( y )   Q m    u C u (11)
dy dt dy
Rearranging, we get a 2nd order non-homogeneous ordinary differential equation.
d 2Tu ( y ) ds (t ) dTu ( y )
ku 2
  u Cu   b C b wbTu    b C b wbTb  Q m (12)
dy dt dy
The corresponding boundary conditions are given as,
T f ( y   s (t ), t )  T0 (13a)
Tu ( y  , t )  Ti (13b)
Tu ( y  0, t )  T f ( y  0, t )  Tm (13c)

The solution of Eq. (10) can be obtained by integrating twice, yields

C1  s y 
T f ( y)  exp    C2 (14)
 s   
f  f 

𝑑𝑠(𝑡) 𝑘𝑓
where, 𝑠 ′ = 𝑑𝑡
, and 𝛼𝑓 = 𝜌 . Here 𝐶1 and 𝐶2 are arbitrary constants.
𝑓 𝐶𝑓

The solution of Eq. (12) is obtained as

 b Cb wbTb  Q m
Tu ( y)  C3 exp1 y   C 4 exp2 y   (15)
 b Cb wb
−𝜌𝑢 𝐶𝑢 𝑠′ −√(𝜌𝑢 𝐶𝑢 𝑠′ )2 +4𝑘𝑢 𝐶𝑏 𝑤𝑏 −𝜌𝑢 𝐶𝑢 𝑠′ +√(𝜌𝑢 𝐶𝑢 𝑠′ )2 +4𝑘𝑢 𝐶𝑏 𝑤𝑏
where, 𝜆1 = and 𝜆2 = are roots of the
2𝑘𝑢 2𝑘𝑢
characteristic equation. Here 𝐶3 and 𝐶4 are arbitrary constants.

2.2 Solution to the Stefan Problem

Since the temperature of the solid and liquid phases is taken to be equal to the clearly defined melting
temperature, this problem is also known as the single interface or isothermal phase change problem. The
solution to the Stefan problem was obtained by substituting the value of blood perfusion and metabolic
generation equal to zero in Eq. (12).

4
 TFEC-2020-36282

The value of arbitrary constants 𝐶1 and 𝐶2 are obtained by substituting the boundary conditions (Eqs
13a and c) into Eq. (14):
s Tm  T0 
C1  (16a)
f
 ss  
exp  1
 
 f 
Tm  T0 
C 2  Tm  (16b)
 ss  
exp  1
 
 f 

The temperature distribution inside the unfrozen region can be expressed as

𝑠′
exp(− 𝑦)
𝛼𝑢
𝑇𝑢 (𝑦) = 𝐶3 𝑠′
+ 𝐶4 (17)
(− )
𝛼𝑢
𝑠′
where, 𝐶3 = (𝑇𝑖 − 𝑇𝑚 ) ( ) and 𝐶4 = 𝑇𝑖
𝛼𝑢

The values of constant 𝐶1 and 𝐶2 , and 𝐶3 and 𝐶4 have been substituted in Eqs. (14) and (17) for the
temperature distribution inside the frozen and unfrozen region, respectively.
In the solid or frozen domain,
 x  s (t ) ds (t ) 
1  exp   
  dt 
T f ( x, t )  Tm  (Ti  Tm )  f  (18)
 s (t ) ds (t ) 
1  exp  
 dt 
 f 
In the liquid or unfrozen domain,
  x  s (t ) ds (t )  
Tu ( x, t )  Tm  (Ti  Tm )  1  exp    (19)
   dt 
  f 

𝑑𝑠(𝑡)
It is to be noted here that the interface position (𝑠(𝑡)) and velocity ( 𝑑𝑡 ) are unknown in Eqs. (18) and (19).
Therefore, the temperature fields for the frozen and unfrozen domain have been substituted into Eq. (4d) to
determine the interface positions, and the equation is given below:

𝑠′
𝑘𝑓 (𝑇𝑜 −𝑇𝑚 )( )
𝛼𝑓 𝑠′
− 𝑘𝑢 (𝑇𝑖 − 𝑇𝑚 ) (𝛼 ) = 𝜌𝐿𝑠 ′ (20)
𝑠𝑠′ 𝑢
1−𝑒𝑥𝑝( )
𝛼𝑓

After rearranging the terms in Eq. (20), we get

𝑑𝑠(𝑡)
𝑠(𝑡) = 𝛼𝑓 ln 𝐶5 (21)
𝑑𝑡
k f T0  Tm 
f
where C5  1 
k u Ti  Tm 
is constant.
f L
u
The above equation is integrated and apply the condition 𝑠 (𝑡 = 0) = 0, we get the interface positions as a
function of time (Eq. 22).

5
 TFEC-2020-36282

s(t )  2 f ln( C5 )t (22)

It is found that the interface position is the function of the square root of time t.

The results were then plotted using MATLAB to give the Temperature vs. Position, Interface Position
variation, and Temperature vs. Time graphs.

2.3 Solution to the Pennes Bioheat Equation

The primary concern of this study is the phase change heat transfer process in biological tissue during
cryosurgery. Hence a solution to the bio-heat transfer equation has been developed using the Coordinate
Transformation technique (described earlier in Section 2.1). The temperature distribution inside the biological
tissue has been obtained by finding the constants in Eqs. (14) and (15) using the boundary conditions (13). The
final solution thus becomes
In the frozen domain:
   ys  
1  exp  f  
T f ( y )  Tm  (Tm  T0 ) 
 (23)
 1  exp ss  
   f  
In the unfrozen domain:
Tu ( y )  Ti  (Ti  Tm )  e1 y (24)

For the interface position (𝑠(𝑡)), we get the following equation in the same way as discussed in Section 2.2:

 k f Tm  T0 s
 ku 1 Ti  Tm    f L
ds(t )
(25)
  
 f 1  exp ss
dt
 
  f 
Eq. (25) can’t be solved by direct analytical methods, and hence, the exponential term is expanded into series
form till the fifth power. The resulting non-linear, implicit ODE is then solved numerically using MATLAB’s
ode15i solver. The result obtained for the variation of the interface position is then plugged into Eqs. (23) and
(24) to get the temperature distributions.

The thermo-physical properties for biological tissue are given in Table 1.

Table 1 The thermo-physical properties of soft biological tissue [8, 11]

Parameter Symbol Value Units
Density of frozen tissue ρf 921 kg/m3
Density of unfrozen tissue ρu 1050 kg/m3
Density of blood ρb 1000 kg/m3
Specific heat of the frozen tissue Cf 1800 J/kg∙K
Specific heat of unfrozen tissue Cu 3770 J/kg∙K
Specific heat of blood Cb 3600 J/kg∙K
Thermal conductivity of frozen tissue kf 2 W/m∙K
Thermal conductivity of unfrozen tissue ku 0.5 W/m∙K
Latent Heat L 250 kJ/kg
Blood temperature Tb 310.15 K
Body core temperature Ti 310.15 K
Phase change temperature Tm 272.15 K
Cryoprobe temperature T0 77 K
Blood perfusion rate ρbwb 0.5 kg/m3-s
Metabolic heat generation rate Qm 4200 W/m3

6
 TFEC-2020-36282

3. RESULTS AND DISCUSSION

The results obtained using the analytical solution (presented in Section 2) have been discussed in this section. First,
the results of the Stefan problem has been verified with results available in [12]. Subsequently, the effect of Stefan
number on the interface position and temperature distribution inside the biological tissue has been discussed. Once
the solutions have been verified with the published results, the biological tissue results during cryosurgery are
discussed.

3.1 Verification of analytical solution

The mathematical model developed for the 1-D phase change heat transfer problem was verified using the
exact solution given in [12]. Both the interface position variation and the temperature distribution at a particular
time instant have been compared to establish the accuracy of the developed solution. The Interface position
variation with respect to time is shown in Fig. 1. As shown in the figure, there is a very good agreement
between the developed and solutions in [12]. Next, we verify the temperature distribution plot at t = 100 s. The
spatial variation of temperature distribution at t = 100 s is shown in Fig. 2 and good agreement as well.
However, there is a slight variation, especially in the solid or frozen region. While the temperature predicted
by the current work is slightly above that predicted by the solution from [12], the trend reverses in the
liquid/unfrozen region.

Fig. 1 Comparison of interface position variation of present results with results given in [12].

Fig. 2 Comparison of temeprature distribution of present results with results given in [12] at t = 100 s.

7
 TFEC-2020-36282

3.2 Effect of Stefan number

A parametric study was conducted to understand the effect of change in various physical factors on the
temperature distribution and the interface position variation. The Stefan number, which is defined as the ratio
of the sensible heat to the latent heat in any phase change heat transfer process and it is defined as,
C p T
Ste  (26)
L
From Eq. (26), it is evident that the Stefan number changes when the thermal conductivity, temperature
difference, or latent heat changes. Hence, the latent heat was varied to vary the Stefan number.

The variation of the interface position with respect to time is shown in Fig. 3 for varying Stefan number. As
the Stefan number decreases, the interface propagation slows down. This can be attributed to the fact that a
low Stefan number corresponds to greater latent heat compared to sensible heat, meaning the phase change
requires greater energy and thereby time. Fig. 4 depicts that for a higher Stefan number, the domain has cooled
down more, and also, the phase change interface has propagated further for higher value of Stefan number.
This is also attributed to the fact that lower values of the Stefan number correspond to greater latent heat
compared to sensible heat. Hence the process of phase change and consequently interface propagation are
slower.

Fig. 3 Variation of the Interface position for different Stefan number.

Fig. 4 Variation of Temperature distribution for different Stefan number.

8
 TFEC-2020-36282

3.3 Pennes Bioheat Equation

The temperature distribution is shown in Fig. 5 at different time instants. There is approximately linear
temperature variation in the frozen domain from the cryoprobe temperature to the tissue freezing temperature.
In contrast, the temperature in the unfrozen domain rises to equal the core body temperature. The extent of the
frozen region increases with time, and the frozen region's temperature also continuously drops with time. Fig.
6 shows the variation of temperature with time at different distances from the cryoprobe surface. At x = 5 mm
and x = 10 mm, the stark change in temperature gradient above and below the freezing temperature is seen
clearly. This is because of higher heat flux in the unfrozen region attributed to the metabolic heat generation,
latent heat generation as well as different thermal properties on either side of the interface.

Fig. 5 The spatial temperature distribution at different time instants.

Fig. 6 Temperature variation with time at different positions

The technique developed in the present study easily traces the interface position at any time instants. In contrast,
the transcendental equation for the interface positions is given in [12], which takes more computational time
to get the interface positions at any instants than the in-house developed technique.

9
 TFEC-2020-36282

4. CONCLUSIONS

The novel technique for the phase change problem in biological tissue has been developed using the coordinate
transformation method. The analytical solution of the freezing/ thawing of biological tissue has been presented
in this study. The results obtained using the Stefan problem's analytical solution are verified with results
available in the literature and found that they match each other reasonably. The effect of Stefan number on
interface position and temperature distribution are studied. The developed analytical solution correctly models
the freezing of biological tissue during cryosurgery. The computational time for solving the governing equation
is almost negligible, which can enable widespread adoption of simulation in aiding surgeons for planning
cryosurgery.

NOMENCLATURE

L Latent Heat of Fusion (kJ/kg) s Interface position (m)

T0 Cryoprobe Temperature (K) u Unfrozen (subscript) (-)
Tb Temperature of blood (K) f Frozen (subscript) (-)
Tm Melting Temperature (K) b Blood (subscript) (-)
Ti Core body Temperature (K) wb Blood perfusion rate (kg/s/ml)
Qm Metabolic heat generation (W/m3)

REFERENCES
[1] Ahmadikia, H., Moradi, “Non-Fourier phase change heat transfer in biological tissues during solidification”, Heat Mass Transf.,
48, pp. 1559-1568, (2012).
[2] Kumar, M., Upadhyay, S., Rai, K.N., “A study of heat transfer during cryosurgery of lung cancer”, J. Therm. Biol., 84, pp. 53-
73, (2019).
[3] Baust, J.G., Gage, A.A., Bjerklund Johansen, T.E., Baust, J.M., “Mechanisms of cryoablation: Clinical consequences on
malignant tumors”, Cryobiology, 68, pp.1-11, (2014).
[4] Fortin, A, Belhamadia, Y,“Numerical prediction of freezing fronts in cryosurgery: comparison with experimental results”,
Comput. Meth. Biomech. Biomed. Eng., 8 (4), pp. 241-249, (2005).
[5] Kudryashov, N.A, Shilnikov, K.E., “Numerical modeling and optimization of the cryosurgery operations”, J. Comp. Appl. Math.,
290, pp. 259-267, (2015).
[6] Deng, Z.-S., Liu, J., “Numerical simulation of 3-D freezing and heating problems for combined cryosurgery and hyperthermia
therapy”, Numer. Heat Transf. A, 46, pp. 587-611, (2004).
[7] Kumar, A., Kumar, S., Katiyar, V.K., Telles, S., “Phase change heat transfer during cryosurgery of lung ancer using hyperbolic
heat conduction model”, Comput. Biol. Med., 84, pp. 20-29, (2017).
[8] Singh, S., Kumar, S., “Numerical study on triple layer skin tissue freezing using dual phase lag bio-heat model”, Int. J. Therm.
Sci., 86, pp. 12-20, (2014).
[9] Mochnacki, B., Majchrzak, E., “Numerical model of thermal interactions between cylindrical cryoprobe and biological tissue
using the dual-phase lag equation”, Int. J. Heat Mass Transf., 108, pp. 1-10, (2017).
[10] Ahmadikia, H., Moradi, A., “Non-Fourier phase change heat transfer in biological tissues during solidification”, Heat Mass
Transf., 48, pp. 1559-1568, (2012).
[11] Moradi, A., Ahmadikia, H., “Numerical study of the solidification process in biological tissue with blood flow and metabolism
effects by the dual phase lag model”, Proc IMech Part H: J. Eng. Med., 226(5), pp. 406-416, (2012).
[12] Hahn, D.W., Ozisik, M.N., Heat conduction, John Wiley & Sons, pp. 452–607 (2012).

10