1 Abstract— A two-dimensional temperature estimation method is proposed based on the detection of shifts in speckle pattern location of ultrasound B-Mode digital images from a region of tissue undergoing thermal therapy. The speckle pattern shifts are due to the local temperature dependence of sound speed and thermal expansion in the heated region. Here in this paper we make use of optical-flow in order to achieve a better accuracy and lower computational burden. The accuracy of measurement of the temperature has been tested on simulated images and it is experimentally validated using tissue mimicking phantom. Good agreement was obtained from the simulated sequences (mean difference= 0.1 and pick error was 0.6). The proposed method is found computationally faster compared to our previous work which was based on the cross correlation algorithm. I. INTRODUCTION inimally Invasive Thermal Therapy such as Intensity Focused Ultrasound (HIFU)[1], Laser Induced Thermal Therapy (LITT) [2] and etc have gained increasing attention in the last decade as alternatives to the standard surgical therapy. Thermal monitoring is one of the important factors that are necessary during the thermo therapy treatment, to reduce treatment imposition and prevent of necrosis in the normal tissue. Lack of efficient, noninvasive and fast method to estimate the tissue temperature causes the reduction of acceptability of this treatment, in spite of its capability in soft tissue cancer treatment. Many researchers have been trying to develop new Manuscript received 17, April, 2008 Bahram M. Mehrabani with the Department of Biomedical Systems & Medical Physics, School of Medicine, Medical Sciences/University of Tehran and Research Center for Science and Technology in Medicine, RCSTIM, Tehran, Iran. (email: Bmomen@razi.tums.ac.ir ) Vahid Tavakoli with the Univercity of Louisville, Louisville, KY, USA. ( email: V0tava01@louisville.eu ) Mohammad D. Abolhassani is Associate Professor of Medical Physics and Biomedical Engineering Department of Tehran University of Medical Sciences and Head of Biomedical Group of Research Centre For Science & Technology in Medicine Tehran University of Medical Science, Iran. (E-mail: abolhasm@sina.tums.ac.ir ) Javad Alirezaie, Senior Member IEEE with the Dept. Electrical Engineering, Ryerson University, Toronto, Canada. (email: javad@ee.ryerson.ca ) Alireza Ahmadian, Senior Member IEEE, with the Department of Biomedical Systems & Medical Physics, School of Medicine, Medical Sciences/University of Tehran and Research Center for Science and Technology in Medicine, RCSTIM, Tehran, Iran. (email: Ahmadian@sina.tums.ac.ir ) quantitative non-invasive techniques for the tissue temperature measurement and treatment monitoring. These techniques are based on Magnetic Resonance Imaging (MRI) [1], Impedance tomography, Microwave Radiometry, and Ultrasound. The noninvasive thermal monitoring using ultrasound waves is the most compatible method in this treatment especially when HIFU method is used, because of three major reasons: First, the nature of treatment and recognition devices are the same. Second the ultrasound Imaging can develop real time image of the tissue. Third, in commercial HIFU systems, a diagnostic ultrasound probe has been mounted in the center of coaxial high intensity ring probe to generate image of the region undergoing the heating and specify the tumor depth and the depth that HIFU must be focused. Therefore, no other diagnostic device is needed for estimating the temperature map of the tissue. Among the ultrasound techniques, different approaches have been suggested to estimate the temperature changes: analysis of the frequency dependent attenuation [2], backscattered power [3], and speed of sound and thermal expansion [4-6]. By increasing the temperature of the region, the local speed of sound in that region will increase and a time-shift in RF-echo will appear. This is called virtual shift. Another effect causing a real shift in the location of the backscatters is the thermal expansion. Most of researchers have implemented the system based on the RF backscatter signal to estimate the time-shift but this signal is classified as the raw data that is not accessible in the commercial systems. In our previous work [7], we used simple speckle tracking methods in digital ultrasonic images instead of using RF backscatter signal. The main drawback of this method was the computational cost of the correlation algorithm due to using block based approach. In this continuation work we have improved our method based on the optical flow to make the algorithm faster and more efficient. II. MATERIAL AND METHODS A Theory: The sound speed in the most of organs increases, when the temperature in that region rises. The biological tissues can be modeled with respect to some facts: An Efficient Temperature Estimation Using Optical-Flow in Ultrasound B-Mode Digital Images Bahram M. Mehrabani, Vahid Tavakoli, Mohammad D. Abolhassani, Member IEEE, Javad Alirezaie, Senior Member IEEE, Alireza Ahmadian, Senior Member IEEE M