- Email: [email protected]
Basic characteristics of the radio imaging method for biomedical applications
Medical Engineering & Physics 26 (2004) 431–437 www.elsevier.com/locate/medengphy
Communication
Basic characteristics of the radio imaging method for biomedical applications I. Hieda a,, K.C. Nam b, A. Takahashi a a
Neuromuscular Technology Group, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 6, 1-1-1 Higashi, Tsukuba 305-8566, Ibaraki, Japan b Graduate Program in Biomedical Engineering, Yonsei University, Seoul, South Korea Received 23 January 2003; received in revised form 13 January 2004; accepted 16 February 2004
Abstract The radio imaging method (RIM), a technology used practically for geophysical surveys, was applied to biomedical measurements. The characteristics of a subject between a pair of simple loop antennas were measured by using a feeble electromagnetic wave of low frequency. Water distribution inside the human body was expected to be measured, as well as the original method imaged mineral distributions. In geophysical surveys, finer resolution is provided than a wavelength of the electromagnetic wave. This was also expected for biomedical measurements. An acrylic water tank was used as a phantom of the abdominal portion of the human body, and basic measurements were performed with the phantom. The inner tank of the phantom, whose cross section was 6 8 cm, was detected by the method when the frequency was 54 MHz or the wavelength was nearly 6 m in free space. The resolution provided by the experiment suggested that the proposed method was effective even if the wavelength was longer than the dimensions of the targeted area. The advantage of the method is simplicity, economy and safety. The authors are looking for specific applications for example, a urination sensor or a vital monitor for in-home care, where the strong points of the method are suitable. # 2004 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Radio imaging method; RIM; Biomedical measurement; Electromagnetic wave; VHF; Magnetic induction tomography
1. Introduction We are developing biomedical measurement technology by using electromagnetic waves of relatively low frequency. The method is based on the radio imaging method (RIM) which is used practically in the mining industry [16,18]. In a successful example [20], the frequency of the electromagnetic wave used for the measurement was 50–520 kHz. The wavelength of the frequency in free space is approximately 600 m–6 km. Small loop antennas, whose diameter was 48 mm, were used for both transmission and reception. A pair of antennas was placed at a maximum of 100 m separation. The measured electric field intensity was assumed to be the integral of the variable, which Corresponding author. Tel.: +81-29-861-9430; fax: +81-29-8616660. E-mail address: [email protected] (I. Hieda).
related to loss and dielectric permittivity, along the straight line from the transmitting antenna to the receiving antenna. Because the wavelength was longer than the dimensions of the subject, the resolution integrated in the measured data had not been expected to be very high. However, a resolution of several tens of meters, which was noticeably finer than the wavelength, was given by a devised expert system for an inverse problem [12,20]. When RIM is applied to biomedical measurement, the scale of the subjects is remarkably different from that of geological survey subjects using the original method. The wavelength should be shortened to balance with the dimension of the subjects. According to literature citations [6,9], penetration of an electromagnetic wave in to the living human body deteriorates when the frequency is higher than 1 GHz. Our experiments started at 1.2 GHz, or a wavelength of 25 cm, expecting a resolution of several centimeters, by using a
1350-4533/$ - see front matter # 2004 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2004.02.005
432
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Fig. 1. Overview of the measurement: subject passes between a pair of antennas and the change of signal strength is measured.
living human abdomen and an acrylic tank filled with water as shown in Fig. 1 [4]. The signal strength at the receiving side was not enough because attenuation by the human abdomen was significant at this frequency range. Reproducibility of the experiments was not satisfactory either. The measurement inaccuracy was caused by interaction of reflected signals from the measurement environment. Introducing an electromagnetic darkroom causes loss of simplicity, one of the main advantages of the method. A pilot study was repeated with water tanks and the abdominal region by lowering the frequency. At a 50 MHz frequency range, stable and reproducible measurement results were obtained. There are several studies of biomedical measurement, which use electromagnetic waves. One example is microwave tomography [8,13]. The basic principle of those projects is similar to our proposal. There are also trials to measure temperature distribution in the human body with a radar system [10,11]. The devices for these projects become large-scaled because they use antennas of high efficiency and sharp directivity, aiming for high resolution. Because attenuation in the human body at the microwave frequency range is remarkable, an electromagnetic darkroom is necessary to prevent interference of electromagnetic waves along indirect paths. The expected resolution of the methods was not worth the cost. One of the features of biomedical measurement by RIM is simplicity. The expected cost of a practical device is as low as that of other simple measurements methods for example, the bio-impedance method [1,3,19] and ultrasonic scanners. Though weak electromagnetic waves in the VHF range penetrate the living human body, invasiveness of and influence on subjects is trivial or comparable to that of mobile phones. Electromagnetic waves pass through cavities and bones without attenuation, which is the main artifact of the bio-impedance method and ultrasonic scanners. The
precision and resolution of the measurements are expected to exceed those of a bio-impedance measurement. Another advantage is that the proposed method does not need direct contact with the skin of the subject by electrodes. Contact resistance between electrode and skin often causes inaccuracy and affects reproducibility in bioelectric measurements like EMG, EEG and the bio-impedance method [5]. The proposed method can avoid not only the causes of inaccuracy but also stimulation of the skin. The antenna used for the proposed method is a simple loop antenna; it can be made as a spiral loop antenna, which is as thin as the electrode used for bioelectric measurements. Measurement is also possible when the subject is fully clothed. It could be applied as a urine volume monitoring system in the bladder and in vital sensors of respiration and blood flow that are used on a daily basis. Development of CT targeting cerebral blood flow distribution is also planned. The aims of this paper are to introduce a biomedical application of the RIM and to show the possibility of its practical use by discussing the results of our preliminary experiments. 2. Method A water tank, whose dimensions were comparable to the abdomen of a human was used for quantitative measurement. The water tank had a dual structure and was made of 5 mm thick acrylic board as shown in Fig. 2. All dimensions in the figure represent interior measurements in millimeters. The height describes the standard water level of the experiments. The inner tank simulated the large internal organs of a human for example, the bladder. The volume of the inner tank was 1.2 l which was larger than human bladders; however, the cross section of the inner tank in the horizontal plane is comparable to that of a human bladder.
Fig. 2. Schematic figure of the water tank: the tank is made of 5 mm thick acrylic board. The numerical values indicate inner dimensions.
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Because this basic study limited the structure to two dimensions, the height of the inner tank was not considered. The measurement environment was a study room in a reinforced concrete building without special electric or magnetic shielding. According to observations before and during experiments, noise and unexpected electromagnetic waves that could influence an experiment were not measured in the room. The structure of an antenna is shown in Fig. 3. The same type of antenna was used for both transmission and reception. Because the diameter of the antenna was significantly smaller than the wavelength, it was difficult to make a stable antenna that was precisely resonant to the frequency. The antenna illustrated in the figure was a magnetic driven antenna that was not resonant to a specific frequency in the range of the targeting frequencies. A forced balun (balance–unbalance transformer) was inserted at the feed point to prevent unexpected radiation and reception from a coaxial cable whose characteristic impedance was 50 X [14]. The power of the transmission signal was 0.4 W and the frequency was 54 MHz. The signal received on the receiving antenna was fed to a spectrum analyzer, Anritsu MS2621B. The spectrum analyzer simply measured the signal strength of the fundamental harmonic of the received signal. At the same time, noise level and unwanted electromagnetic waves were monitored. The transmitting and receiving antennas and the water tank were arranged so that the heights of the centers were 90 cm from the floor. Three different separations of the pair of antennas, 100, 50 and 30 cm, were chosen as experimental conditions. The water tank was moved by 2 cm to go across the centerline of the transmitting and receiving antennas at the center point, and signal strength was measured at every step. For each antenna spacing, four conditions of the water
Fig. 3. Schematic figure of a loop antenna used for the experiment: a balance–unbalance transformer (balun) made of a toroidal core is placed at the feed point to prevent unexpected transmission and reception by the coaxial cable.
433
tank were set: both inner and outer tanks empty, innertank-filled, outer tank filled and both tanks filled. Tap water for drinking was used for the experiment because tap water, distilled water and aqueous salt solutions of various densities had given no differences of signal strength under the same conditions in a preparatory experiment. Twenty measurements were performed under each condition, and the measured signal strength was averaged. 3. Results and discussion Fig. 4a–c shows the results of the mean signal strengths of 20 measurements when the distance of the transmitting and receiving antennas was 30, 50 and 100 cm, respectively. Each figure shows the measured signal strengths where both inner and outer tanks were empty, inner tank was filled, outer tank was filled and both tanks were filled. The reference signal strength was set to a 2.66 mV input voltage by the spectrum analyzer. During pilot studies prior to the experiment described in this paper, the reference had been chosen by considering the margin to noise and dynamic range of the measurement apparatus. However, the relation between the reference level and absolute electric field intensity of the receiving antenna was not confirmed. When the outer tank was filled or both tanks were filled, the signal strength curves had gentle peak shapes as shown in Fig. 4a,b. Attenuation of the signal was expected when the water tank was placed between the two antennas because of eddy current loss and other losses. However, the signal strengths increased approximately 3 dB in Fig. 4a and 2 dB in Fig. 4b when the water tank was located across the two antennas. The influence of diffraction could be ignored when the distance between the antennas was greater than k=2p, where k is the wavelength. Under this condition, loss caused by DC resistance was dominant and attenuation of signal strength was considered as the integration of the losses along the line between the two antennas [17]. In Fig. 4c, when the antenna distance was 100 cm, the distance was greater than k=2p, or 88 cm at 54 MHz in free space, and no significant change in the signal strength was discovered. In Fig. 4a,b, where the antenna distances were 30 and 50 cm, respectively, both distances were less than k=2p in free space. The relative permittivity of water compared to free space is approximately 70 at room temperature in the VHF frequency range including 54 MHz, though the permittivity depends on the temperature and pressure [3]. The wavelength was 10 cm in water or living tissue that contains a relatively high level of moisture. The distances of 30 and 50 cm were greater than k=2p if the wavelength in water and living tissue was considered. However, the signal strength curves had a gentle peak
434
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Fig. 4. Results of the measurement: vertical axis: averaged signal strength (db) of 20 measurements; horizontal axis: water tank location. Water tank is moved in 2 cm steps. Antenna distance (a) 30 cm; (b) 50 cm; (c) 100 cm. In each figure, dotted, broken, and solid lines correspond to conditions where the inner tank is filled, the outer tank is filled and both tanks are filled, respectively. The alternate long and short dashed line represents the result of the empty tank. Dots on the horizontal axis indicate the points where significant difference is detected by two-sided t-test between outer tank and both tank cases at a significance level of 0.01.
shape unlike the square one that originated from the water tank. These curves indicated that the measurement was influenced by diffraction. The loss of electromagnetic waves in water or living tissue that contains moisture is not remarkable in the VHF range including 54 MHz [9], and Fig. 4a,b showed that the electric field intensity was increased by
the permittivity and that its effect was more influential than the expected losses. In Fig. 4a,b, the differences in the peaks of the outer tank and both tanks were approximately 0.3 and 0.1 dB, respectively. Two-sided t-tests were performed between the outer tank and both of the tanks for all the measurement points in the two figures. Dots on the horizontal axes indicated points where significant differences were detected at the significance level of 0.01. The points were concentrated at the center, where the inner tank was located. In Fig. 4a, where the antenna separation was 30 cm, these points spread through 20 cm,which was greater than the width of the inner tank, 8 cm. The results of the t-test suggest that the measurement was effective even if the influence of diffraction was remarkable when the antenna separation was 50 cm or less. By lowering the frequency and using small loops, the proposed method had similarity to magnetic induction tomography (MIT) in the methodologies. In MIT, a coil was driven by a current of 10 kHz through 10 MHz, and the conductivity of the subject was derived from the measured voltage and phase on the other coil [2,7,15]. Because the mutual inductance reduced remarkably, the distance between the coil pair had limitation. On the other hand, our proposed method successfully detected significant differences caused by the inner cavity of the water tank when the distances between the antennas were 30 and 50 cm. The distances were larger than those of most MIT studies, and they were suitable for measurement of the living human body. As shown in Fig. 4a,b, the signal strengths of the inner tank produced small peaks. Not only was the peak level significantly smaller than the peaks of the outer tank and both tanks, but the shape of the peak was also sharper. In Fig. 4a, the signal strength of the empty tank had small peaks which are remarkably smaller than the peak of the inner tank. The peaks were provided at positions 12 and 23 where the sidewalls of the outer tank crossed the antennas. The relative permittivity of acrylic at the frequency range was 3.0, smaller than that of water. Moreover, the thickness of the sidewalls was 5 mm,which was small enough in comparison with the dimension of the tank. The figure shows that effect of acrylic walls on the electric field intensity was not remarkable. In Fig. 4b,c, where the distances of the antennas were 50 and 100 cm, respectively, the effect of the empty tank was not recognized. Fig. 5 shows a comparison of the signal strengths of the inner tank and subtraction of the outer tank from both tanks. To compensate for the effect of acrylic walls, the curves were also plotted where the signal strengths of the empty tank were subtracted from those of the inner tank. The axis of the signal strength is a
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Fig. 5. Signal strength of the inner tank and differences of both tanks from the outer tank. Vertical axis: signal strength deviation in linear scale; horizontal axis: water tank location. Antenna distance (a) 30 cm; (b) 50 cm. The alternate long and short dashed line represents inner-tank-filled response compensated for the empty tank response.
linear scale, different from the dB scales in Fig. 4. When the separation of antennas was 30 cm, the difference of the outer tank from both tanks agreed with that of the inner tank. The peak signal strength was reduced, after compensating for the empty tank. The peak levels of the three curves deviated a little but the deviations were not remarkable. The curve ‘‘both—outer’’ should agree with the curve ‘‘inner—empty’’ because the contributions of the acrylic tank were eliminated in the both curves. The shapes and the widths of the two peaks matched with each other; however, the leading and trailing portions of the peak had different shapes. Some effect was recognized by the empty tank compensation method. For 50 cm antenna spacing, it was not clear whether the three patterns did or did not have relationships because the influence of external disturbances was significant. These results show that the addition and subtraction of signal strengths measured by this method can be performed on a linear scale when the distance between
435
the antennas is 30 cm. This is one of the important conditions for a two- or three-dimensional inverse problem to build an image of the inside of a body. In the experiment, only water, one of the components of living body was used as the phantom medium. The maximum change of signal strength was 3 dB. However, the change occurred when the inside and outside of the tank were filled with water, whose relative permittivity is 70. Signal strength difference was only 3% whether the inner tank was full or empty, as shown in Fig. 5a.The permittivities of the compositions of a living human body except water were generally much smaller than that of water. When the effects of an electromagnetic wave on a living body were considered, the tissues were classified into two groups: components containing a high level of moisture and those with a low level of moisture. Therefore, the experiments and the analysis with the phantom were primarily performed to simulate components containing a high level of moisture. If the proposed method worked ideally without diffraction effect or other artifacts, the measured signal strengths should change in proportion to the thickness of the water across the two antennas as illustrated in Fig. 6, in spite of the difference in the antenna distance. The effects of the acrylic walls were ignored for simplicity. For both tanks and for the inner tank, the ideal shapes became trapezoids whose dimensions depended on their cross sections. When only the outer tank was filled, the shape was more complicated. The upper side of the trapezoid included a dimple in the shape of a small trapezoid. The actual signal strengths shown in Fig. 4 were regarded as outputs of the ideal shapes shown in Fig. 6 that passed through space as a low pass filter that is, the measurement system. In Fig. 4a–c, the
Fig. 6. Water tank function, that is based on the thickness of the water column between the transmission and receiving antennas. Vertical axis: signal strength (no unit); horizontal axis: water tank location. The thickness and influence of the tank wall is ignored for simplification. Dotted, broken and solid lines correspond to conditions where the inner tank is filled, the outer tank is filled and both tanks are filled, respectively.
436
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
signal strengths, when the inner tank was filled, were less than those of both tanks but the patterns did not have a dimple. The resolution of the method was greater than the 2 cm of the moving step of the experiment it did not seem enough for detection of the inner tank. The spatial frequency response of Fig. 4a is the basis for the following discussion concerning minimum external disturbances. Fig. 7a shows the frequency responses of the raw signals or patterns in Fig. 6 given by FFT. The horizontal and vertical axes describe the spatial frequency and signal strength, respectively. Both axes have logarithmic scales. The three polygonal lines are the signal strengths of the raw signals in spatial frequency, and they correspond to the inner tank, the outer tank and both tanks from the bottom to the top of the figure, respectively. The simple lines accompanying the polygonal lines are their respective regression lines given by the least square method. The slopes of the three simple lines are the same because they have ideal frequency responses. Fig. 7b represents the spatial frequency response of Fig. 4a when the antenna separation is 30 cm. The sim-
Fig. 7. Signal strength of (a) water tank function (Fig. 5) and (b) measured value (Fig. 4a) in the spatial frequency domain. Both horizontal and vertical axes are logarithmic scales. Polygonal lines are frequency responses given by FFT, and linear lines accompanying them are the respective regression lines given by the least square method.
ple regression lines are also given by the least square method for the polygonal lines. It is obvious that the high frequency components of the outer tank and both of the tanks are significantly reduced. It can be said, as a general tendency, that the measurement worked as a low pass filter in the spatial frequency domain. On the other hand, the slope of the line for the inner tank data was the same as that in Fig. 7a. The high frequency components had not been lost. This suggests that the proposed method provided resolution as fine as 2 cm under some conditions related to the distance from the antennas. To construct the CT image, the influences of diffraction and distortion caused by the speed differences of electromagnetic waves in water or tissue should be compensated for. An interactive or expert system, which refers to the electric geometry database of the human body, should be considered. 4. Conclusions The proposed method used VHF electromagnetic waves whose wavelength was remarkably longer than the dimensions of the human body in free space. The resolution, in general, tended to be not fine enough, but the possibility to provide better resolution that was fine enough to image internal organs was also indicated. The addition and subtraction of results that were required for an inverse problem was also shown. We plan further experiments to refine the relation of the resolution and spacing of antennas and distances of material and antennas. Improving antenna characteristics and finding a more suitable frequency is also necessary. Resolutions of 1 cm to several centimeters are expected. Analytical study of antennas and media is indispensable to reveal the theoretical limit and optimum conditions of the measurement. Because of the methodological similarity, a combination of MIT and the proposed method would be possible and would be effective in providing higher level information on the living human body. Depending on the final resolution of measurement, the methods can be applied to monitor urine volume inside the bladder, edema of limbs, function of lungs and cerebral blood flow. Such monitors will be appropriate for the home care of patients and the handicapped, where safety, ease of use, and the economical aspect are important.
References [1] Brown BH. Impedance tomography and spectroscopy: what can and what will we see? Proceedings of the XI International Conference on Electrical Bio-Impedance. 2001, p. 9–13.
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437 [2] Griffiths H, Steward WR, Gough W. Magnetic induction tomography: a measuring system for biological tissues. Annals of the New York Academy of Sciences 1999;873:335–45. [3] Grimnes S, Martinsen OG. Bioimpedance and bioelectricity basics. Academic Press; 2000. [4] Hieda I. Preliminary study of non-contact bio-impedance measurement by RIM. Proceedings of the XI International Conference on Electrical Bio-Impedance. 2001, p. 645–8. [5] Hieda I, Kuchinomachi Y, Sato S, Kodama H. Impedance mapping method and system for blood flow measurement in brain. Journal of Robotics and Mechatronics 1992;4:314–20. [6] Johnson CC, Guy AW. Nonionizing electromagnetic wave effects in biological materials and systems. Proceedings of the IEEE 1972;60:692–718. [7] Korjenevsky A, Cheripenin V, Sapetsky S. Magnetic induction tomography: experimental realization. Physiological Measurement 2000;21(1):89–94. [8] Larsen LE, Jacobi JH. Medical applications of microwave imaging. IEEE Press; 1986. [9] Miyakawa M. Noninvasive measurement of temperature profiles inside dielectric materials. Bulletin of the Electrotechnical Laboratory 1982;45:419–35. [10] Miyakawa M. Tomographic measurement of temperature change in phantoms of the human body by chirp radar-type microwave computed tomography. Medical and Biological Engineering Computing 1993;31(Suppl):S31–6. [11] Miyakawa M, Orikasa K, Bertero M, Boccacci P, Conte F, Piana M. Experimental validation of a linear model for data
[12]
[13] [14] [15]
[16]
[17]
[18]
[19] [20]
437
reduction in chirp-pulse microwave CT. IEEE Transactions on Medical Imaging 2002;21:385–95. Rogers G, Brandt L, Young J, Kot J. The study of diffusion effects in RIM tomographic imaging. Exploration Geophysics 1993;24:785–8. Rosen A, Rosen HD. New frontiers in medical device technology. New York: John Wiley & Sons; 1995. Sevick J. Transmission line transformers. Georgia: Noble Publishing; 2001. Scharfetty H, Lackner HK, Rosell J. Magnetic induction tomography: hardware for multifrequency measurements in biological tissue. Physiological Measurement 2001;22(1): 131–146. Stolarczyk LG, Fry RC, Lloyd TW. Radio imaging method (RIM) used to map coal seam thickness within developed longwall panels. AAPG Bulletin 1984;68:951. Stolarczyk LG. Definition imaging of an orebody with the radio imaging method (RIM). IEEE Transactions on Industry Applications 1992;28:1141–7. Thomson S, Young J, Sheard N. Base metal application of the radio imaging method: current status and case studies. Exploration Geophysics 1992;23:367–72. Webster JG, editor. Electrical impedance tomography. Bristol and New York: Adam Hilger; 1990. Young J, Rogers G, Thomson S, Neil M. Australian development of tomographic radio imaging as a new tool in mining geophysics. Butsuri-Tansa 1994;47:249–55.
Communication
Basic characteristics of the radio imaging method for biomedical applications I. Hieda a,, K.C. Nam b, A. Takahashi a a
Neuromuscular Technology Group, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 6, 1-1-1 Higashi, Tsukuba 305-8566, Ibaraki, Japan b Graduate Program in Biomedical Engineering, Yonsei University, Seoul, South Korea Received 23 January 2003; received in revised form 13 January 2004; accepted 16 February 2004
Abstract The radio imaging method (RIM), a technology used practically for geophysical surveys, was applied to biomedical measurements. The characteristics of a subject between a pair of simple loop antennas were measured by using a feeble electromagnetic wave of low frequency. Water distribution inside the human body was expected to be measured, as well as the original method imaged mineral distributions. In geophysical surveys, finer resolution is provided than a wavelength of the electromagnetic wave. This was also expected for biomedical measurements. An acrylic water tank was used as a phantom of the abdominal portion of the human body, and basic measurements were performed with the phantom. The inner tank of the phantom, whose cross section was 6 8 cm, was detected by the method when the frequency was 54 MHz or the wavelength was nearly 6 m in free space. The resolution provided by the experiment suggested that the proposed method was effective even if the wavelength was longer than the dimensions of the targeted area. The advantage of the method is simplicity, economy and safety. The authors are looking for specific applications for example, a urination sensor or a vital monitor for in-home care, where the strong points of the method are suitable. # 2004 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Radio imaging method; RIM; Biomedical measurement; Electromagnetic wave; VHF; Magnetic induction tomography
1. Introduction We are developing biomedical measurement technology by using electromagnetic waves of relatively low frequency. The method is based on the radio imaging method (RIM) which is used practically in the mining industry [16,18]. In a successful example [20], the frequency of the electromagnetic wave used for the measurement was 50–520 kHz. The wavelength of the frequency in free space is approximately 600 m–6 km. Small loop antennas, whose diameter was 48 mm, were used for both transmission and reception. A pair of antennas was placed at a maximum of 100 m separation. The measured electric field intensity was assumed to be the integral of the variable, which Corresponding author. Tel.: +81-29-861-9430; fax: +81-29-8616660. E-mail address: [email protected] (I. Hieda).
related to loss and dielectric permittivity, along the straight line from the transmitting antenna to the receiving antenna. Because the wavelength was longer than the dimensions of the subject, the resolution integrated in the measured data had not been expected to be very high. However, a resolution of several tens of meters, which was noticeably finer than the wavelength, was given by a devised expert system for an inverse problem [12,20]. When RIM is applied to biomedical measurement, the scale of the subjects is remarkably different from that of geological survey subjects using the original method. The wavelength should be shortened to balance with the dimension of the subjects. According to literature citations [6,9], penetration of an electromagnetic wave in to the living human body deteriorates when the frequency is higher than 1 GHz. Our experiments started at 1.2 GHz, or a wavelength of 25 cm, expecting a resolution of several centimeters, by using a
1350-4533/$ - see front matter # 2004 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2004.02.005
432
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Fig. 1. Overview of the measurement: subject passes between a pair of antennas and the change of signal strength is measured.
living human abdomen and an acrylic tank filled with water as shown in Fig. 1 [4]. The signal strength at the receiving side was not enough because attenuation by the human abdomen was significant at this frequency range. Reproducibility of the experiments was not satisfactory either. The measurement inaccuracy was caused by interaction of reflected signals from the measurement environment. Introducing an electromagnetic darkroom causes loss of simplicity, one of the main advantages of the method. A pilot study was repeated with water tanks and the abdominal region by lowering the frequency. At a 50 MHz frequency range, stable and reproducible measurement results were obtained. There are several studies of biomedical measurement, which use electromagnetic waves. One example is microwave tomography [8,13]. The basic principle of those projects is similar to our proposal. There are also trials to measure temperature distribution in the human body with a radar system [10,11]. The devices for these projects become large-scaled because they use antennas of high efficiency and sharp directivity, aiming for high resolution. Because attenuation in the human body at the microwave frequency range is remarkable, an electromagnetic darkroom is necessary to prevent interference of electromagnetic waves along indirect paths. The expected resolution of the methods was not worth the cost. One of the features of biomedical measurement by RIM is simplicity. The expected cost of a practical device is as low as that of other simple measurements methods for example, the bio-impedance method [1,3,19] and ultrasonic scanners. Though weak electromagnetic waves in the VHF range penetrate the living human body, invasiveness of and influence on subjects is trivial or comparable to that of mobile phones. Electromagnetic waves pass through cavities and bones without attenuation, which is the main artifact of the bio-impedance method and ultrasonic scanners. The
precision and resolution of the measurements are expected to exceed those of a bio-impedance measurement. Another advantage is that the proposed method does not need direct contact with the skin of the subject by electrodes. Contact resistance between electrode and skin often causes inaccuracy and affects reproducibility in bioelectric measurements like EMG, EEG and the bio-impedance method [5]. The proposed method can avoid not only the causes of inaccuracy but also stimulation of the skin. The antenna used for the proposed method is a simple loop antenna; it can be made as a spiral loop antenna, which is as thin as the electrode used for bioelectric measurements. Measurement is also possible when the subject is fully clothed. It could be applied as a urine volume monitoring system in the bladder and in vital sensors of respiration and blood flow that are used on a daily basis. Development of CT targeting cerebral blood flow distribution is also planned. The aims of this paper are to introduce a biomedical application of the RIM and to show the possibility of its practical use by discussing the results of our preliminary experiments. 2. Method A water tank, whose dimensions were comparable to the abdomen of a human was used for quantitative measurement. The water tank had a dual structure and was made of 5 mm thick acrylic board as shown in Fig. 2. All dimensions in the figure represent interior measurements in millimeters. The height describes the standard water level of the experiments. The inner tank simulated the large internal organs of a human for example, the bladder. The volume of the inner tank was 1.2 l which was larger than human bladders; however, the cross section of the inner tank in the horizontal plane is comparable to that of a human bladder.
Fig. 2. Schematic figure of the water tank: the tank is made of 5 mm thick acrylic board. The numerical values indicate inner dimensions.
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Because this basic study limited the structure to two dimensions, the height of the inner tank was not considered. The measurement environment was a study room in a reinforced concrete building without special electric or magnetic shielding. According to observations before and during experiments, noise and unexpected electromagnetic waves that could influence an experiment were not measured in the room. The structure of an antenna is shown in Fig. 3. The same type of antenna was used for both transmission and reception. Because the diameter of the antenna was significantly smaller than the wavelength, it was difficult to make a stable antenna that was precisely resonant to the frequency. The antenna illustrated in the figure was a magnetic driven antenna that was not resonant to a specific frequency in the range of the targeting frequencies. A forced balun (balance–unbalance transformer) was inserted at the feed point to prevent unexpected radiation and reception from a coaxial cable whose characteristic impedance was 50 X [14]. The power of the transmission signal was 0.4 W and the frequency was 54 MHz. The signal received on the receiving antenna was fed to a spectrum analyzer, Anritsu MS2621B. The spectrum analyzer simply measured the signal strength of the fundamental harmonic of the received signal. At the same time, noise level and unwanted electromagnetic waves were monitored. The transmitting and receiving antennas and the water tank were arranged so that the heights of the centers were 90 cm from the floor. Three different separations of the pair of antennas, 100, 50 and 30 cm, were chosen as experimental conditions. The water tank was moved by 2 cm to go across the centerline of the transmitting and receiving antennas at the center point, and signal strength was measured at every step. For each antenna spacing, four conditions of the water
Fig. 3. Schematic figure of a loop antenna used for the experiment: a balance–unbalance transformer (balun) made of a toroidal core is placed at the feed point to prevent unexpected transmission and reception by the coaxial cable.
433
tank were set: both inner and outer tanks empty, innertank-filled, outer tank filled and both tanks filled. Tap water for drinking was used for the experiment because tap water, distilled water and aqueous salt solutions of various densities had given no differences of signal strength under the same conditions in a preparatory experiment. Twenty measurements were performed under each condition, and the measured signal strength was averaged. 3. Results and discussion Fig. 4a–c shows the results of the mean signal strengths of 20 measurements when the distance of the transmitting and receiving antennas was 30, 50 and 100 cm, respectively. Each figure shows the measured signal strengths where both inner and outer tanks were empty, inner tank was filled, outer tank was filled and both tanks were filled. The reference signal strength was set to a 2.66 mV input voltage by the spectrum analyzer. During pilot studies prior to the experiment described in this paper, the reference had been chosen by considering the margin to noise and dynamic range of the measurement apparatus. However, the relation between the reference level and absolute electric field intensity of the receiving antenna was not confirmed. When the outer tank was filled or both tanks were filled, the signal strength curves had gentle peak shapes as shown in Fig. 4a,b. Attenuation of the signal was expected when the water tank was placed between the two antennas because of eddy current loss and other losses. However, the signal strengths increased approximately 3 dB in Fig. 4a and 2 dB in Fig. 4b when the water tank was located across the two antennas. The influence of diffraction could be ignored when the distance between the antennas was greater than k=2p, where k is the wavelength. Under this condition, loss caused by DC resistance was dominant and attenuation of signal strength was considered as the integration of the losses along the line between the two antennas [17]. In Fig. 4c, when the antenna distance was 100 cm, the distance was greater than k=2p, or 88 cm at 54 MHz in free space, and no significant change in the signal strength was discovered. In Fig. 4a,b, where the antenna distances were 30 and 50 cm, respectively, both distances were less than k=2p in free space. The relative permittivity of water compared to free space is approximately 70 at room temperature in the VHF frequency range including 54 MHz, though the permittivity depends on the temperature and pressure [3]. The wavelength was 10 cm in water or living tissue that contains a relatively high level of moisture. The distances of 30 and 50 cm were greater than k=2p if the wavelength in water and living tissue was considered. However, the signal strength curves had a gentle peak
434
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Fig. 4. Results of the measurement: vertical axis: averaged signal strength (db) of 20 measurements; horizontal axis: water tank location. Water tank is moved in 2 cm steps. Antenna distance (a) 30 cm; (b) 50 cm; (c) 100 cm. In each figure, dotted, broken, and solid lines correspond to conditions where the inner tank is filled, the outer tank is filled and both tanks are filled, respectively. The alternate long and short dashed line represents the result of the empty tank. Dots on the horizontal axis indicate the points where significant difference is detected by two-sided t-test between outer tank and both tank cases at a significance level of 0.01.
shape unlike the square one that originated from the water tank. These curves indicated that the measurement was influenced by diffraction. The loss of electromagnetic waves in water or living tissue that contains moisture is not remarkable in the VHF range including 54 MHz [9], and Fig. 4a,b showed that the electric field intensity was increased by
the permittivity and that its effect was more influential than the expected losses. In Fig. 4a,b, the differences in the peaks of the outer tank and both tanks were approximately 0.3 and 0.1 dB, respectively. Two-sided t-tests were performed between the outer tank and both of the tanks for all the measurement points in the two figures. Dots on the horizontal axes indicated points where significant differences were detected at the significance level of 0.01. The points were concentrated at the center, where the inner tank was located. In Fig. 4a, where the antenna separation was 30 cm, these points spread through 20 cm,which was greater than the width of the inner tank, 8 cm. The results of the t-test suggest that the measurement was effective even if the influence of diffraction was remarkable when the antenna separation was 50 cm or less. By lowering the frequency and using small loops, the proposed method had similarity to magnetic induction tomography (MIT) in the methodologies. In MIT, a coil was driven by a current of 10 kHz through 10 MHz, and the conductivity of the subject was derived from the measured voltage and phase on the other coil [2,7,15]. Because the mutual inductance reduced remarkably, the distance between the coil pair had limitation. On the other hand, our proposed method successfully detected significant differences caused by the inner cavity of the water tank when the distances between the antennas were 30 and 50 cm. The distances were larger than those of most MIT studies, and they were suitable for measurement of the living human body. As shown in Fig. 4a,b, the signal strengths of the inner tank produced small peaks. Not only was the peak level significantly smaller than the peaks of the outer tank and both tanks, but the shape of the peak was also sharper. In Fig. 4a, the signal strength of the empty tank had small peaks which are remarkably smaller than the peak of the inner tank. The peaks were provided at positions 12 and 23 where the sidewalls of the outer tank crossed the antennas. The relative permittivity of acrylic at the frequency range was 3.0, smaller than that of water. Moreover, the thickness of the sidewalls was 5 mm,which was small enough in comparison with the dimension of the tank. The figure shows that effect of acrylic walls on the electric field intensity was not remarkable. In Fig. 4b,c, where the distances of the antennas were 50 and 100 cm, respectively, the effect of the empty tank was not recognized. Fig. 5 shows a comparison of the signal strengths of the inner tank and subtraction of the outer tank from both tanks. To compensate for the effect of acrylic walls, the curves were also plotted where the signal strengths of the empty tank were subtracted from those of the inner tank. The axis of the signal strength is a
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
Fig. 5. Signal strength of the inner tank and differences of both tanks from the outer tank. Vertical axis: signal strength deviation in linear scale; horizontal axis: water tank location. Antenna distance (a) 30 cm; (b) 50 cm. The alternate long and short dashed line represents inner-tank-filled response compensated for the empty tank response.
linear scale, different from the dB scales in Fig. 4. When the separation of antennas was 30 cm, the difference of the outer tank from both tanks agreed with that of the inner tank. The peak signal strength was reduced, after compensating for the empty tank. The peak levels of the three curves deviated a little but the deviations were not remarkable. The curve ‘‘both—outer’’ should agree with the curve ‘‘inner—empty’’ because the contributions of the acrylic tank were eliminated in the both curves. The shapes and the widths of the two peaks matched with each other; however, the leading and trailing portions of the peak had different shapes. Some effect was recognized by the empty tank compensation method. For 50 cm antenna spacing, it was not clear whether the three patterns did or did not have relationships because the influence of external disturbances was significant. These results show that the addition and subtraction of signal strengths measured by this method can be performed on a linear scale when the distance between
435
the antennas is 30 cm. This is one of the important conditions for a two- or three-dimensional inverse problem to build an image of the inside of a body. In the experiment, only water, one of the components of living body was used as the phantom medium. The maximum change of signal strength was 3 dB. However, the change occurred when the inside and outside of the tank were filled with water, whose relative permittivity is 70. Signal strength difference was only 3% whether the inner tank was full or empty, as shown in Fig. 5a.The permittivities of the compositions of a living human body except water were generally much smaller than that of water. When the effects of an electromagnetic wave on a living body were considered, the tissues were classified into two groups: components containing a high level of moisture and those with a low level of moisture. Therefore, the experiments and the analysis with the phantom were primarily performed to simulate components containing a high level of moisture. If the proposed method worked ideally without diffraction effect or other artifacts, the measured signal strengths should change in proportion to the thickness of the water across the two antennas as illustrated in Fig. 6, in spite of the difference in the antenna distance. The effects of the acrylic walls were ignored for simplicity. For both tanks and for the inner tank, the ideal shapes became trapezoids whose dimensions depended on their cross sections. When only the outer tank was filled, the shape was more complicated. The upper side of the trapezoid included a dimple in the shape of a small trapezoid. The actual signal strengths shown in Fig. 4 were regarded as outputs of the ideal shapes shown in Fig. 6 that passed through space as a low pass filter that is, the measurement system. In Fig. 4a–c, the
Fig. 6. Water tank function, that is based on the thickness of the water column between the transmission and receiving antennas. Vertical axis: signal strength (no unit); horizontal axis: water tank location. The thickness and influence of the tank wall is ignored for simplification. Dotted, broken and solid lines correspond to conditions where the inner tank is filled, the outer tank is filled and both tanks are filled, respectively.
436
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437
signal strengths, when the inner tank was filled, were less than those of both tanks but the patterns did not have a dimple. The resolution of the method was greater than the 2 cm of the moving step of the experiment it did not seem enough for detection of the inner tank. The spatial frequency response of Fig. 4a is the basis for the following discussion concerning minimum external disturbances. Fig. 7a shows the frequency responses of the raw signals or patterns in Fig. 6 given by FFT. The horizontal and vertical axes describe the spatial frequency and signal strength, respectively. Both axes have logarithmic scales. The three polygonal lines are the signal strengths of the raw signals in spatial frequency, and they correspond to the inner tank, the outer tank and both tanks from the bottom to the top of the figure, respectively. The simple lines accompanying the polygonal lines are their respective regression lines given by the least square method. The slopes of the three simple lines are the same because they have ideal frequency responses. Fig. 7b represents the spatial frequency response of Fig. 4a when the antenna separation is 30 cm. The sim-
Fig. 7. Signal strength of (a) water tank function (Fig. 5) and (b) measured value (Fig. 4a) in the spatial frequency domain. Both horizontal and vertical axes are logarithmic scales. Polygonal lines are frequency responses given by FFT, and linear lines accompanying them are the respective regression lines given by the least square method.
ple regression lines are also given by the least square method for the polygonal lines. It is obvious that the high frequency components of the outer tank and both of the tanks are significantly reduced. It can be said, as a general tendency, that the measurement worked as a low pass filter in the spatial frequency domain. On the other hand, the slope of the line for the inner tank data was the same as that in Fig. 7a. The high frequency components had not been lost. This suggests that the proposed method provided resolution as fine as 2 cm under some conditions related to the distance from the antennas. To construct the CT image, the influences of diffraction and distortion caused by the speed differences of electromagnetic waves in water or tissue should be compensated for. An interactive or expert system, which refers to the electric geometry database of the human body, should be considered. 4. Conclusions The proposed method used VHF electromagnetic waves whose wavelength was remarkably longer than the dimensions of the human body in free space. The resolution, in general, tended to be not fine enough, but the possibility to provide better resolution that was fine enough to image internal organs was also indicated. The addition and subtraction of results that were required for an inverse problem was also shown. We plan further experiments to refine the relation of the resolution and spacing of antennas and distances of material and antennas. Improving antenna characteristics and finding a more suitable frequency is also necessary. Resolutions of 1 cm to several centimeters are expected. Analytical study of antennas and media is indispensable to reveal the theoretical limit and optimum conditions of the measurement. Because of the methodological similarity, a combination of MIT and the proposed method would be possible and would be effective in providing higher level information on the living human body. Depending on the final resolution of measurement, the methods can be applied to monitor urine volume inside the bladder, edema of limbs, function of lungs and cerebral blood flow. Such monitors will be appropriate for the home care of patients and the handicapped, where safety, ease of use, and the economical aspect are important.
References [1] Brown BH. Impedance tomography and spectroscopy: what can and what will we see? Proceedings of the XI International Conference on Electrical Bio-Impedance. 2001, p. 9–13.
I. Hieda et al. / Medical Engineering & Physics 26 (2004) 431–437 [2] Griffiths H, Steward WR, Gough W. Magnetic induction tomography: a measuring system for biological tissues. Annals of the New York Academy of Sciences 1999;873:335–45. [3] Grimnes S, Martinsen OG. Bioimpedance and bioelectricity basics. Academic Press; 2000. [4] Hieda I. Preliminary study of non-contact bio-impedance measurement by RIM. Proceedings of the XI International Conference on Electrical Bio-Impedance. 2001, p. 645–8. [5] Hieda I, Kuchinomachi Y, Sato S, Kodama H. Impedance mapping method and system for blood flow measurement in brain. Journal of Robotics and Mechatronics 1992;4:314–20. [6] Johnson CC, Guy AW. Nonionizing electromagnetic wave effects in biological materials and systems. Proceedings of the IEEE 1972;60:692–718. [7] Korjenevsky A, Cheripenin V, Sapetsky S. Magnetic induction tomography: experimental realization. Physiological Measurement 2000;21(1):89–94. [8] Larsen LE, Jacobi JH. Medical applications of microwave imaging. IEEE Press; 1986. [9] Miyakawa M. Noninvasive measurement of temperature profiles inside dielectric materials. Bulletin of the Electrotechnical Laboratory 1982;45:419–35. [10] Miyakawa M. Tomographic measurement of temperature change in phantoms of the human body by chirp radar-type microwave computed tomography. Medical and Biological Engineering Computing 1993;31(Suppl):S31–6. [11] Miyakawa M, Orikasa K, Bertero M, Boccacci P, Conte F, Piana M. Experimental validation of a linear model for data
[12]
[13] [14] [15]
[16]
[17]
[18]
[19] [20]
437
reduction in chirp-pulse microwave CT. IEEE Transactions on Medical Imaging 2002;21:385–95. Rogers G, Brandt L, Young J, Kot J. The study of diffusion effects in RIM tomographic imaging. Exploration Geophysics 1993;24:785–8. Rosen A, Rosen HD. New frontiers in medical device technology. New York: John Wiley & Sons; 1995. Sevick J. Transmission line transformers. Georgia: Noble Publishing; 2001. Scharfetty H, Lackner HK, Rosell J. Magnetic induction tomography: hardware for multifrequency measurements in biological tissue. Physiological Measurement 2001;22(1): 131–146. Stolarczyk LG, Fry RC, Lloyd TW. Radio imaging method (RIM) used to map coal seam thickness within developed longwall panels. AAPG Bulletin 1984;68:951. Stolarczyk LG. Definition imaging of an orebody with the radio imaging method (RIM). IEEE Transactions on Industry Applications 1992;28:1141–7. Thomson S, Young J, Sheard N. Base metal application of the radio imaging method: current status and case studies. Exploration Geophysics 1992;23:367–72. Webster JG, editor. Electrical impedance tomography. Bristol and New York: Adam Hilger; 1990. Young J, Rogers G, Thomson S, Neil M. Australian development of tomographic radio imaging as a new tool in mining geophysics. Butsuri-Tansa 1994;47:249–55.