# Image-based computational model for focused ultrasound ablation of liver tumor

- Maxim A Solovchuk
^{1}Email author, - Tony WH Sheu
^{1, 2}Email author and - Marc Thiriet
^{3}

**1**:4

https://doi.org/10.1186/2194-3990-1-4

© Solovchuk et al.; licensee Springer. 2014

**Received: **25 March 2013

**Accepted: **9 October 2013

**Published: **10 January 2014

## Abstract

### Abstract

High-intensity focused ultrasound (HIFU) is a rapidly developing medical technology that allows non-invasive thermal ablation of tumors. Thermal treatment of liver tumor, which is one of the most common malignancies worldwide, is problematic because large blood vessels act as a heat sink. Convective cooling protects the cancer cells from thermal destruction and decreases the necrosed volume. A major objective of the method development is to achieve a virtually complete necrosis of tumors close to major blood vessels and to avoid blood vessel damage and, hence, the needed treatment planning. The present study is aimed at predicting liver tumor temperature during HIFU thermal ablation in a patient-specific liver geometry. The model comprises the nonlinear Westervelt equation and bioheat equations in the liver and blood vessels. The nonlinear hemodynamic equations are also taken into account with the convected cooling and acoustic streaming effects being taken into account. We found from this three-dimensional three-field coupling study that in large blood vessels, both convective cooling and acoustic streaming may change the temperature considerably near the blood vessel. More precisely, acoustic streaming velocity magnitude can be several times larger than the blood vessel velocity. The results presented in the current work can be further used to construct a surgical planning platform.

### Keywords

HIFU Liver tumor Blood vessel Westervelt equation Acoustic streaming.## Background

High-intensity focused ultrasound (HIFU) is a therapeutic method for a non-invasive ablation of benign and malignant tumors [1, 2]. The main mechanism of tissue ablation is thermal coagulation. Temperature about 60°C can cause an irreversible tissue damage. Liver cancer is the second leading cause of death in Asia [3] and is now known as one of the leading causes of death in the world. The primary problem in the thermal ablation therapy of liver tumor is due to a heat sink resulting from the blood flow in large blood vessels.

Convective cooling protects tumor cells from thermal destruction [4, 5] and, consequently, causes the recurrent cancer. Blood flow can carry away a large part of the deposited energy; therefore, more energy is necessary for the ablation of tumor close to blood vessel. This can lead to the use of redundant ultrasound powers and undesirable damage of healthy tissues. Special care should be taken to avoid destruction of vessel walls by a high temperature. In some studies it was reported that high-intensity focused ultrasound can cause occlusion of blood vessels and blood vessel disruption [6, 7]. A basic understanding of the factors that can altogether influence the tissue necrosis volume is necessary to improve the thermoablative therapy and prevent recurrence.

One clinical trial in [8] showed that HIFU can safely necrotize the tumors close to major hepatic veins. Before HIFU treatment, some patients had transcatheter chemoembolization. After a single session of HIFU treatment, the rate of complete necrosis was about 50%, which is not satisfactory at all. Lack of a complete response can be attributed to the large tumor size and the cooling effect in large vessels.

Liver is a highly perfused organ with a blood supply from the hepatic artery and portal vein. In most of the previous computational studies, liver is considered as a homogeneous tissue, and the amount of dissipated heat is usually estimated by averaging the effect of blood perfusion over all tissues [9]. This approach is valid for tissues with capillaries. However, for tissues with thermally significant blood vessels (diameters larger than 0.5 mm) the biologically relevant convective cooling needs to be taken into account [10], and homogenization assumption is no longer valid. Kolios et al. [10] and Curra et al. [11] studied the influence of blood vessels on the lesion size based on a 2D finite difference calculation in the cylindrical coordinate system. The focal point was at the center of the blood vessel. In this case, a large portion of the deposited ultrasound energy was carried away by convective cooling. Recently, a three-dimensional (3D) model to determine the influence of blood flow on the temperature distribution was presented [12, 13]. To get the velocity distribution for a real blood vessel geometry, nonlinear hemodynamic equations were used. Focused ultrasound beam can induce an additional mass flow in blood vessels. This effect is known as an acoustic streaming. Acoustic streaming can increase the velocity magnitude in a blood vessel, strengthen the heat sink, and change the temperature distribution near the blood vessel [13]. Therefore, the effect of acoustic streaming was taken into account in the present study.

In the present work, temperature elevation in the liver tumor is considered in a patient-specific geometry. Tumor is located close to the portal vein with the diameter about 7 mm. The theoretical feasibility to ablate tumor close to the portal vein was shown. The results presented in the current work can be further used to construct a surgical planning platform. The computational model can play an important role in the planning and optimization of the treatment based on the patient’s image.

## Methods

### Nonlinear acoustic equation

In the above, *p* is the sound pressure, $\beta =1+\frac{B}{2A}$ the coefficient of nonlinearity, and *δ* the diffusivity of sound. The first two terms describe the linear lossless wave propagating at a small-signal sound speed. The third term represents the loss due to thermal conduction and fluid viscosity. The last term accounts for acoustic nonlinearity which may considerably affect thermal and mechanical changes within the tissue.

where *α*_{abs} denotes the acoustic absorption coefficient.

### Energy equation for tissue heating

Whereas hepatic arteries and portal veins irrigate the liver parenchyma, hepatic veins drain blood out of the liver and can thus be considered as a heat sink. Tumor cells in perivascular region, as a result, may escape from an externally imposed large heat, leading possibly to a local recurrence. Therefore, the mathematical model suitable for predicting the temperature in tissues must take the heat conduction, tissue perfusion, convective blood cooling, and heat deposition due to an incident wave into account. In the simulation of the thermal field, the physical domain has been split into the domains for the perfused tissue and the flowing blood.

In the above bioheat equation proposed for modeling the time-varying temperature in tissue domain, *ρ*, *c*, and *k* denote the density, specific heat, and thermal conductivity, respectively, with the subscripts *t* and *b* referring to the tissue and blood domains, respectively. The notation *T*_{
∞
} is denoted as the temperature at a remote location. The variable *w*_{
b
} (≡10 kg/m^{3} s) in Equation 4 is the perfusion rate for the tissue cooling in capillary flows. It is noted that the above bioheat equation for *T* is coupled with the Westervelt equation (1) for the acoustic pressure through a power deposition term *q* defined in Equation 3.

*q*, and the heat sink, which is $-{\rho}_{b}\phantom{\rule{2.22144pt}{0ex}}{c}_{b}\phantom{\rule{2.22144pt}{0ex}}\underline{\mathbf{u}}\xb7\nabla T$, are added to the conventional diffusion-type heat equation. The resulting energy equation given below avoids a possible high recurrence stemming from the tumor cell survival next to large vessels

In the above equation, $\underline{\mathbf{u}}$ is the blood flow velocity. Owing to the presence of blood flow velocity vector $\underline{\mathbf{u}}$ in the energy equation, a biologically sound model needed for conducting a HIFU simulation should comprise a coupled system of thermal-fluid-acoustics nonlinear differential equations.

where *R*=2 for *T*>=43°C, *R*=4 for 37°C<*T*<43°C. The value of TD required for a total necrosis ranges from 25 to 240 min in biological tissues [13, 17, 18]. According to this relation, thermal dose resulting from heating the tissue to 43°C for 240 min is equivalent to that achieved by heating it to 56°C for 1 s.

### Acoustic streaming hydrodynamic equations

**P**is the static pressure,

*μ*(=0.0035 kg/m s) the shear viscosity of blood flow, and

*ρ*the blood density. In Equation 7, the force vector $\underline{\mathbf{F}}$ acting on the blood fluid due to ultrasound is assumed to be along the acoustic axis $\underline{\mathbf{n}}$. The resulting nonzero component in $\underline{\mathbf{F}}$ takes the following form [20]:

### Three-point sixth-order accurate scheme for Westervelt equation

*u*

^{(2)}≡

*u*

_{ x x },

*u*

^{(4)}≡

*u*

_{ x x x x }and

*u*

^{(6)}at a nodal point

*j*as follows:

*x*

_{ j }is to relate

*t*,

*s*, and

*w*with

*u*as follows:

By substituting the Taylor series expansion into Equation 14 and then conducting a term-by-term comparison of the derivatives, the introduced free parameters can be determined as *α*_{1}=*α*_{-1}=-1, *α*_{0}=2, *β*_{0}=-1, ${\gamma}_{0}=-\frac{1}{12}$, and ${\delta}_{0}=-\frac{1}{360}$.

*s*

_{ j }=

*k*

_{ j }

*u*

_{ j }+

*f*

_{ j }, the following two expressions for ${t}_{j}=({k}_{j}^{2}{u}_{j}+2{k}_{x,j}{u}_{x,j}+{k}_{\mathit{\text{xx}},j}{u}_{j}+{k}_{j}\phantom{\rule{0.3em}{0ex}}{f}_{j}+{f}_{\mathit{\text{xx}},j})$ and ${w}_{j}=({k}_{j}^{3}{u}_{j}+7{k}_{j}{u}_{j}{k}_{\mathit{\text{xx}},j}+6{k}_{j}{u}_{x,j}{k}_{x,j}+4{k}_{x,j}^{2}{u}_{j}+6{k}_{\mathit{\text{xx}},j}\phantom{\rule{0.3em}{0ex}}{f}_{j}+4{k}_{\mathit{\text{xxx}},j}{u}_{x,j}+{k}_{\mathit{\text{xxxx}},j}{u}_{j}+4{k}_{x,j}\phantom{\rule{0.3em}{0ex}}{f}_{x,j}+{k}_{j}\phantom{\rule{0.3em}{0ex}}{f}_{\mathit{\text{xx}},j}+{f}_{\mathit{\text{xxxx}},j})$ are resulted. Equation 14 can then be rewritten as

where HOT denotes higher order terms. The above modified equation analysis sheds light that the Helmholtz scheme developed within the three-point stencil framework can yield a spatial accuracy order of six. Axisymmetric sound beam was considered. The Westervelt equation was solved in conjunction with the alternating direction implicit solution algorithm [21]. Absorbing boundary conditions were applied to prevent numerical reflections. Accuracy of the numerical solutions was examined [22] by comparing with the known analytical and numerical solutions of other authors.

### Problem description

**Acoustic and thermal properties for the liver tissue and blood**

Tissue | ${\mathbf{c}}_{\mathbf{0}}\mathbf{\left(}\frac{\mathbf{m}}{\mathbf{s}}\mathbf{\right)}$ | $\mathbf{\rho}\left(\frac{\mathit{\text{kg}}}{{\mathbf{m}}^{\mathbf{3}}}\mathbf{\right)}$ | $\mathbf{c}\mathbf{\left(}\frac{\mathbf{J}}{\mathit{\text{kgK}}}\mathbf{\right)}$ | $\mathbf{k}\mathbf{\left(}\frac{\mathbf{W}}{\mathit{\text{mK}}}\mathbf{\right)}$ | $\mathbf{\alpha}\mathbf{\left(}\frac{\mathit{\text{Np}}}{\mathbf{m}}\mathbf{\right)}$ |
---|---|---|---|---|---|

Liver | 1,540 | 1,055 | 3,600 | 0.512 | 8.1 |

Blood | 1,540 | 1,060 | 3,770 | 0.53 | 1.5 |

## Results and discussion

*y*(a) and pressure waveforms at the focal point (b) are presented. At high intensities, nonlinear wave propagation effects lead to the distortion of waveform. Due to the nonlinear distortion, higher harmonics are generated. These higher harmonics are more readily absorbed by the tissue, enhancing therefore local heating. The peak positive and negative pressures are

*P*

_{+}=7.7 MPa and

*P*

_{-}=5.3 MPa, respectively. The linear theory predicts focal pressure

*P*

_{ L }=6.2 MPa. The peak power deposition increased 17% from the linear to nonlinear waveform. The solid tumor was assumed to be exposed to a 1.7-s ultrasound.

*z*=0.16 are presented in Figure 8 at time

*t*=1.7

*s*(end of sonication). Calculation shows that tumor close to the blood vessel can be ablated. There is a very sharp temperature gradient near the blood vessel wall. We can see a strong asymmetry in the lesion shape due to blood flow cooling. The temperature inside the blood vessel remains almost unchanged except the boundary layer close to the focal point. Therefore, focused ultrasound can be a safe therapy to ablate tumors close to the blood vessel wall. However, additional experimental studies

*in vitro*and

*in vivo*are necessary [8]. Simulation results can help to optimize an appropriate sonication time, ultrasound power and focal point location.

**Mass flux**

Outlet 1 | Outlet 2 | Outlet 3 | Outlet 4 | Outlet 5 | Outlet 6 | Inlet | |
---|---|---|---|---|---|---|---|

Without AS | 4.08·10 | 7.59·10 | 6.96·10 | 1.85·10 | 1.615·10 | 1.40·10 | 1.04·10 |

With AS | 3.70·10 | 9.80·10 | 5.07·10 | 2.7·10 | 1.30·10 | 1.22·10 | 1.04·10 |

Difference | -9 | + | -27 | + | -20 | -13 |

The present work demonstrates interference between the parenchymal blood flow and the heat locally transferred to the hepatic parenchyma by ultrasound. On the one hand, the temperature field is affected by blood flow cooling. On the other hand, acoustic streaming resulting from an emitted ultrasound for thermal ablation changes the flow distribution according to the location of large blood vessels with respect to that of the tumor. The closer the vessels to the tumor, the higher the increase of flow rate. Therefore, acoustic streaming contributes to blood flow cooling.

In the present work, the tumor was located close to the blood vessel with the diameter *d*=7 mm. For smaller blood vessel diameters, the effect of acoustic streaming will be smaller. In [24, 25], the importance of blood flow cooling and acoustic streaming was studied for parallel and perpendicular blood vessel orientations. Intermediate intensity range was considered. It was shown that when the distance between the tumor and blood vessel wall is less than several millimeters, the convective cooling should be taken into account and the homogenization assumption becomes no longer acceptable. If the blood vessel is inside the area of the beam, inside the half-pressure maximum (-6 dB) contours, the effect of acoustic streaming should be taken into account. For the case of high intensities, the effect of acoustic streaming becomes important in the larger area. For blood vessels with a smaller diameter, the effect of acoustic streaming becomes less important. It was shown that if the diameter of the blood vessel is smaller than 1 mm acoustic streaming velocity magnitude is smaller than blood flow velocity [24]. However, only intermediate intensity range was considered. For the high-intensity focused ultrasound, acoustic streaming velocity magnitude will be larger and additional parametric studies may be necessary. In the near future, it is highly interesting to investigate the threshold size of blood vessel and the distance between the tumor and blood vessel wall beyond which the acoustic streaming effects should be considered negligibly small in the modeling and calculations.

Personalized liver cancer treatment planning is necessary based on the appropriate image acquisition and processing not only to localize the tumor inside the parenchymal arterial trees and venous drainage network, but also to detect possible small additional tumors more or less close to the main tumor. The existence of these additional foci is the indication for surgical resection.

Post HIFU regrowth depends mainly on the distance between the tumor and large blood vessels. This technique is not used when the distance is too short. Under the circumstances, the surgery is the selected treatment. When the distance is long enough, but remains short, US intensity is limited to avoid a strong vascular damage. The destruction of cancerous cells may then be partial. The post-therapeutic control of the liver must be carefully carried out to ensure that a rapid tumor regrowth does not appear. In the case of tumor reoccurrence, surgery can then be performed. This medical strategy can be used when surgery must be delayed because of the patient state does not enable it. The proper case corresponds to a tumor at distance from large blood vessels, for which the highest admissible dose (US power and duration) can be used.

There are several ways to stop blood flow in portal veins and in hepatic veins either pharmacologically or physically. The proposed mathematical model can be used to predict temperature elevation during the treatment in this case. If blood flow was stopped, it will be easier to ablate tumor close to the blood vessel wall. However, due to a high temperature, the blood vessel wall can be damaged or the blood vessel can be occluded [7]. Prior to the treatment, the model can predict whether the blood vessel will be damaged or not. Depending on the anticipated goals, the current model can be used to optimize ultrasound power and treatment time.

The current work is a first step towards the development of HIFU treatment planning platform. This platform can be used to optimize treatment time and applied ultrasound powers in order to minimize damage of healthy tissues and to destroy all cancer cells.

## Conclusion

The proposed three-dimensional physical model for HIFU study was conducted in an image-based liver geometry. Usually, during the treatment planning, effect of blood flow cooling is not taken into account. In the present work, we showed that both blood flow cooling and acoustic streaming effects can significantly affect the treatment and both of them should be taken into account. At high intensities, the effect of cooling by acoustic streaming can prevail over convective cooling in large blood vessels. The theoretical feasibility to necrotize the tumors close to major veins was shown. It was shown that tumors close to the blood vessel wall can be ablated without damaging the blood vessel wall. These results can be further used to construct a surgical planning platform for the non-invasive HIFU tumor ablating therapy in real liver geometry from MRI image and can lead in the future to the improvement of the focused ultrasound ablation of liver tumor. The presented model can be used to develop similar planning tools for other organs.

## Declarations

### Acknowledgements

The authors would like to acknowledge the financial support from the Center of Advanced Study in Theoretical Sciences (CASTS). A patient-specific liver geometry in Figure 1a was provided by IRCAD (France).

## Authors’ Affiliations

## References

- Zhou Y: High intensity focused ultrasound in clinical tumor ablation.
*World J Clin Oncol*2011, 2: 8–27. 10.5306/wjco.v2.i1.8View ArticleGoogle Scholar - Leslie T, Kennedy J: High intensity focused ultrasound in the treatment of abdominal and gynaecological diseases.
*Int J Hyperthermia*2007, 23(2):173–182. 10.1080/02656730601150514View ArticleGoogle Scholar - Huang JS, Gervais D, Mueller P: Radiofrequency ablation: Review of mechanism, indications, technique, and results.
*Chin J Radiol*2001, 26(3):119–134.Google Scholar - Sheu T, Chou C, Tsai S, Liang P: Three-dimensional analysis for radio-frequency ablation of liver tumor with blood perfusion effect.
*Comput Methods Biomech Biomed Engin*2005, 8(4):229–240. 10.1080/10255840500289731View ArticleGoogle Scholar - Frich L, Hol P, Roy S, Mala T, Edwin B, Clausen O, Gladhaug I: Experimental hepatic radiofrequency ablation using wet electrodes: Electrode-to-vessel distance is a significant predictor for delayed portal vein thrombosis.
*Eur Radiol*2006, 16(9):1990–1999. 10.1007/s00330-006-0177-6View ArticleGoogle Scholar - Wu F, Chen WZ, Bai J, Zou JZ, Wang ZL, Zhu H, Wang ZB: Tumor vessel destruction resulting from high-intensity focused ultrasound in patients with solid malignancies.
*Ultrasound Med Biol*2002, 28(4):535–542. 10.1016/S0301-5629(01)00515-4View ArticleGoogle Scholar - Vaezy S, Zderic V: Hemorrhage control using high intensity focused ultrasound.
*Int J Hyperthermia*2007, 23(2):203–211. 10.1080/02656730601169779View ArticleGoogle Scholar - Zhang L, Zhu H, Jin C, Zhou K, Li K, Su H, Chen W, Bai J, Wang Z: High-intensity focused ultrasound (HIFU): Effective and safe therapy for hepatocellular carcinoma adjacent to major hepatic veins.
*Eur Radiol*2009, 19(2):437–445. 10.1007/s00330-008-1137-0View ArticleGoogle Scholar - Pennes H: Analysis of tissue and arterial blood temperatures in the resting human forearm.
*J Appl Physiol*1948, 1(2):93–122.Google Scholar - Kolios M, Sherar M, Hunt J: Large blood vessel cooling in heated tissues: A numerical study.
*Phys Med Biol*1995, 40(4):477–494. 10.1088/0031-9155/40/4/001View ArticleGoogle Scholar - Curra K, Mourad P, Khokhlova V, Cleveland R, Crum L: Numerical simulations of heating patterns and tissue temperature response due to high-intensity focused ultrasound.
*IEEE Trans Ultrason Ferroelectr Freq Control*2000, 47(4):1077–1089.View ArticleGoogle Scholar - Sheu TW, Solovchuk MA, Chen AW, Thiriet M: On an acoustics-thermal-fluid coupling model for the prediction of temperature elevation in liver tumor.
*Int J Heat Mass Transf*2011, 54(17–18):4117–4126.MATHView ArticleGoogle Scholar - Solovchuk MA, Sheu TWH, Lin WL, Kuo I, Thiriet M: Simulation study on acoustic streaming and convective cooling in blood vessels during a high-intensity focused ultrasound thermal ablation.
*Int J Heat Mass Transf*2012, 55(4):1261–1270. 10.1016/j.ijheatmasstransfer.2011.09.023MATHView ArticleGoogle Scholar - Hamilton MF, Blackstock DT:
*Nonlinear Acoustics*. Boston: Academic; 1998.Google Scholar - Hallaj I, Cleveland R: FDTD simulation of finite-amplitude pressure and temperature fields for biomedical ultrasound.
*J Acoust Soc Am*1999, 105(5):L7-L12. 10.1121/1.426776View ArticleGoogle Scholar - Pierce AD:
*Acoustics: An Introduction to its Physical Principles and Applications*. Melville: Acoustical Society of America; 1991.Google Scholar - Sapareto S, Dewey W: Thermal dose determination in cancer therapy.
*Int J Radiat Oncol Biol Phys*1984, 10(6):787–800. 10.1016/0360-3016(84)90379-1View ArticleGoogle Scholar - Meaney P, Cahill M, Ter Haar G: The intensity dependence of lesion position shift during focused ultrasound surgery.
*Ultrasound Med Biol*2000, 26(3):441–450. 10.1016/S0301-5629(99)00161-1View ArticleGoogle Scholar - Kamakura T, Matsuda K, Kumamoto Y, Breazeale M: Acoustic streaming induced in focused Gaussian beams.
*J Acoust Soc Am*1995, 97(5 I):2740–2746.View ArticleGoogle Scholar - Huang J, Holt R, Cleveland R, Roy R: Experimental validation of a tractable numerical model for focused ultrasound heating in flow-through tissue phantoms.
*J Acoust Soc Am*2004, 116(4 I):2451–2458.View ArticleGoogle Scholar - Sheu T, Hsieh L, Chen C: Development of a three-point sixth-order Helmholtz scheme.
*J Comput Acoust*2008, 16(3):343–359. 10.1142/S0218396X08003610MATHMathSciNetView ArticleGoogle Scholar - Solovchuk MA, Sheu TWH, Thiriet M: Simulation of nonlinear Westervelt equation for the investigation of acoustic streaming and nonlinear propagation effects.
*J Acoust Soc Am*2013, 134(5):3931–3942. 10.1121/1.4821201View ArticleGoogle Scholar - Duck F:
*Physical Property of Tissues - A Comprehensive Reference Book*. London: Academic; 1990.Google Scholar - Solovchuk MA, Sheu TWH, Thiriet M, Lin WL: On a computational study for investigating acoustic streaming and heating during focused ultrasound ablation of liver tumor.
*J Appl Thermal Eng*2013, 56(1–2):62–76.View ArticleGoogle Scholar - Solovchuk MA, Sheu TWH, Thiriet M: The effects of acoustic streaming on the temperature distribution during focused ultrasound therapy.
*AIP Conf Proc*2012, 1433: 589.View ArticleGoogle Scholar

## Copyright

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.