Medical & Biological Engineering & Computing, 1999, 37, pp. 618-624

A catheter tactile sensor for measuring hardness of soft tissue: measurement in a silicone model and in an in vitro human prostate model

A. Eklund¹  A. Bergh²  O.A. Lindahl¹

¹Biomedical Engineering and Informatics, University Hospital, S-901 85, UmeEfont face="Times New Roman">, Sweden

²Department of Pathology, UmeEfont face="Times New Roman"> University, S-901 87 UmeEfont face="Times New Roman">, Sweden

Correspondence should be addressed to A. Eklund;
E-mail: anders.eklund@neuro.umu.se

Abstract - Tissue hardness is related to tissue composition, and this is often changed by disease. It is therefore of interest to measure the hardness in an objective and non-invasive way. A tactile sensor based on a vibrating piezoelectric ceramic element in a feedback loop is described. When the sensor touches an object it produces a frequency shift related to the hardness of the object.  The aim of this study was to develop an in vitro hardness measurement method using a catheter type version of the sensor. The method was evaluated in an established silicone tissue model and on human prostate tissue in vitro. A linear relationship was found with a high degree of explanation (R² = 0.98) between a cone penetration hardness standard (DIN ISO 2137) applied to the silicone model and the corresponding frequency shift.  The results from measurements on a human prostate tissue sample, fixed with formalin, showed that the relative hardness measured with the tactile sensor correlated (R = -0.96, p < 0.001, N = 60) with the proposed hardness related to the histological composition of the prostate tissue.  The results indicated that hardness of prostate tissue, and maybe hardness of human tissue in general, can be expressed according to the cone penetration standard and that the hardness can be measured with this tactile sensory system.   These findings hold the promise of further development of a non-invasive tool for hardness measurement in a clinical situation.

Keywords - Tactile, Sensor, Vibration, Ceramic, Catheter, Hardness, Prostate

List of symbols

1 Introduction

Tissue composition and consistency are often changed by disease. For example, malignant tumours are generally harder than the surrounding tissue, and this is the reason why tumours often can be detected by palpation.  In the female, breast cancers are detected as harder regions imbedded in surrounding normal gland.  Suspicious areas in the breast can be examined further by mammography and biopsy.  In the male, prostate cancer is often detected as a firm nodule in the prostate during rectal palpation. Prostate cancer is the most common cancer in men in the European Union and the USA.  In the US 165,000 men are diagnosed with prostate cancer each year (US Public Health Service, 1993).  Prostate cancer is generally diagnosed by a high blood PSA (prostate specific antigen) level, rectal palpation and ultrasound examination of the prostate followed by histological examination of prostate biopsies. In many patients with high PSA, palpation and ultrasound do not detect any tumour and biopsies are therefore taken at random (Hodge et al., 1989; Aarnink et al., 1998). Therefore, there is a need for improved, non-invasive methods to detect prostate tumours in a reliable and easy way.

Recently a new tactile sensor based on vibration technology for measuring physical properties such as stiffness or hardness of an object was described (Omata and Terunuma, 1992). A ceramic piezoelectric rod is set in oscillation with an electronic feedback circuit.  When the element touches an object the resonance frequency of the oscillating system changes. Preliminary results from measurements of living tissue by detecting the change in frequency have given promising results (Omata and Terunuma, 1991; Lindahl et al., 1998). The tactile sensor technique has been evaluated both in a standardised silicone gum model and in a rat testis model where it was compared with an impression method that measures interstitial pressure and water displacement in ski (Lindahl and Omata, 1995).  It has also been evaluated for detection of changes in stiffness and elastic-related properties of human skin (Lindahl et al., 1998). The results of these studies suggest that the sensor probe with the measuring instrument can measure differences in the stiffness of silicone and may provide information about the properties of skin stiffness and elasticity. Furthermore, a recently published study showed that lymph node stiffness was a useful parameter for diagnosis of metastases in an in vitro setting (Miyagi et al., 1997). Another study with a catheter type tactile sensor in an animal model indicated that direct measurement of bladder stiffness may prove to be a useful tool in the evaluation of bladder and prostate biomechanics (Watanabe et al., 1997).

This study is a first step towards the provision of a non-invasive method for hardness measurement. The study is based on the development of an in vitro hardness measurement method using a catheter type tactile sensor. The method was evaluated in an established silicone tissue model and applied to human prostate tissue in vitro.

2 Materials and Methods

2.1 Tactile sensor theory

The tactile sensor is based on a piezoelectric element shaped like a rod or cylinder and made of lead zirconate titanate (PZT) (Omata and Terunuma, 1992). The element is set in oscillation, close to its fundamental resonance frequency, by means of an electronic feedback circuit. It has a piezoelectric pick-up for detection of the vibrations. This signal is processed in a feedback circuit and drives the PZT-element. The feedback circuit modifies the sinusoidal signal. First, the signal is amplified to a constant amplitude signal so that only the frequency and phase information of the signal are transferred. The signal is then filtered in a bandpass filter to ensure that the PZT-element oscillates in its lowest longitudinal mode. The oscillator frequency is thus solely determined by the zero-phase condition: the sum of the phase shifts around he feedback loop (feedback circuit and PZT-element) must be zero (Floyd, 1988).

When an oscillating rod is applied to an object with a specified acoustic impedance a new oscillating system with a new frequency response characteristic is formed (Omata and Terunuma, 1992). The acoustic impedance of the object, Z, can be separated into a real part, acoustic resistance a, and an imaginary part, acoustic reactance b:

Z = a + jb

(1)

The change in fundamental resonance frequency for a rod of length l may be written as

(2)

where Z₀ is the acoustic impedance of the sensor element and V₀ is the equivalent sound velocity in the sensor element. The reactance can be divided into a mass load part, m, and a stiffness part k:

(3)

where w is the angular frequency and

(4)

(5)

where r is the radius of the contact area, r is density, n is Poisson's ratio and E is Youngs modulus. Depending on the relationship between material constants of the object and the geometry of the probe the sensor system detects mass load or stiffness (Omata and Terunuma, 1992). A decreasing r makes the system more sensitive to k (eqns. 4 and 5). The system output is the shift in the oscillation frequency from the unloaded to the loaded condition (Fig. 1), denoted Df_S. Thus Df_S will be dependent on Z (eqns. 1 and 2), the contact area with the sample (eqns. 4 and 5), the frequency characteristics of the unloaded PZT-element and the frequency characteristics of the feedback circuit through the zero-phase condition.

The visco-elastic behaviour of the sample (Fig. 1) makes it crucial to have a constant time interval, 28-30 seconds, between the moment of contact between object and sensor and the moment of frequency reading.

2.2 Experimental setup

The method was developed and evaluated in two experimental models: first in a silicone model and then in a human prostate model. In common for the two experimental setups were the catheter tip tactile sensor (CTS) and the electronics for the feedback circuit (Fig. 2). The CTS (Prototype by Axiom Co Ltd., Koriyama, Japan) is based on a cylindrical piezoelectric element made of PZT, 7 x 1.2 mm, placed at the end of a catheter, radius r₀ = 1.0 mm. At the tip, in contact with the element, a hemisphere of epoxy is placed, which seals the catheter. Its fundamental resonance frequency is approximately 200 kHz.

Fig. 2 Experimental set-up. (a) Two methods for mounting the sensor were used: for controlled lp application the sensor was mounted on a linear z-translation carrier placed on a stable rod. For constant FC application the sensor was mounted in a counterbalance arrangement. (b) The CTS was driven by the feedback circuit or with a frequency generator. (c) A universal counter was used for phase difference and frequency measurement.

Measurements of frequency and phase differences during the experiments were made with a Universal Counter HP53132 A (Hewlett Packard, Santa Clara, California, USA), with specified resolution 1 Hz and better that 0.1^o, respectively (180-200 kHz range). The vertical z-translation of the CTS was measured with a linear position gauge (Compac, Geneva, Switzerland), type 313K, with resolution of 10 mm. Contact force, F_c, between the sensor tip and the sample was measured with a balance (Mettler-Toledo Inc., Greifensee, Switzerland), PB 602 (Fig. 2), with specified resolution 0.01 g (» 0.1 mN). All measurements were performed at room temperature 22^oC ± 2^oC.

2.3 Silicone model

Silicone rubber has previously been used to model the hardness of living tissue (Lindahl and Omata, 1995). Two-component silicone (Wacker-Chemie GmbH, Munich, Germany), Wacker SilGel 612, was poured into standard Petri dishes, height 15 mm and diameter 87 mm.  The silicone was vulanised into five samples of different hardness, depending on the mixing ratio (Table 1). The hardness of the samples is quoted in the form of penetration values (mm/10) (DIN ISO 2137, 150 g hollow cone), where decreasing penetration values correspond to increasing hardness. Both controlled penetration depth, l_p, and constant F_C set-ups were used on the silicone models. The first set-up aimed to investigate the sensor system's dependency on l_p. The second set-up was used to measure the system's phase-frequency characteristics and to determine a relationship between hardness and frequency shift. The counter was set in average mode (N = 100) and 10 times attenuation on the channel measuring the input to CTS (= feedback-circuit output).

When investigating l_p dependence, the CTS was slowly lowered towards the sample until a minimum force was detected on the balance. The zero position was set at that level and the CTS was then, within approximately 4 seconds, lowered to the chosen l_p. Change in phase shift over CTS, Df_CTS, was measured for l_p = 0.25 mm, 0.5 mm and 1.0 mm. The feedback loop was disconnected during these measurements. Instead a function generator (Hewlett Packard, Santa Clara, California, USA), HP33120A, was used to drive the CTS (Fig. 2). The driving frequency of the sinusoidal wave was 190 kHz and the amplitude was 10 V peak to peak (V_pp), which was comparable to the feedback-circuit output. Phase shift over the CTS was measured before contact and after approximately 30 seconds of penetration.  The difference between the two readings was recorded as Df_CTS. F_C was also measured after approximately 30 seconds of penetration. Ten repeated measurements were made on each l_p for every silicone mixture.

Measurement of phase shifts over the CTS, f_CTS , and over the feedback circuit, f_FC , against frequency, were with a frequency sweep between 187 kHz and 190.5 kHz in step of 100 Hz. f_CTS was recorded for both the unloaded condition and with the tip applied to silicone of mixture 4:3.2 (Table 1). In the latter case the sweep was started after more than 200 seconds of contact and the probe was not moved during the sweep.  For the f_FC measurements the same sweep with a 1.2 V_pp signal was used as input. This was comparable with the output from the CTS. Six repeated sweeps were performed for each condition.

Finally the frequency shift of the sensor system was measured ten times on each of the five silicone samples (Table 1). The feedback circuit was connected to the CTS and the system oscillated at its resonance frequency (Fig. 2). The constant force set-up was used and the frequency was measured with the HP 53132A counter.

2.4 Prostate tissue

For the prostate tissue model the data acquisition was further developed into a computer-controlled set-up with the software developed in LabVIEW (National Instrument, Austin, Texas, USA).  Data transfer to the computer, as well as initialisation and control of the apparatus, was effected using a GPIB interface.   Collected data were presented in real time on the computer screen, the frequency shifts were calculated and the raw data were saved to disk.  All measurements on prostate tissue were made using the constant force method described earlier, and with the feedback circuit connected.  The frequency counter was set in average mode (N = 10).

Prostate tissue was removed from a 72-year-old man suffering from benign prostate hyperplasia (BPH). An approximately 10 mm thick slice, also used for histological diagnosis, was cut and fixed in formalin for 24 h (Tobocman et al., 1997) and then stored in 50% ethanol.  The sample was never kept out of ethanol storage for longer than 30 minutes during measurements, and there was always more than 1 hour between measurement sessions. To map the position of measurements the image of the prostate slice surface was scanned with a flatbed scanner.  The prostate section surface area of the fixed slice was divided by a grid, and six sites were selected for measurement.  To find as homogenous tissue as possible, the measurement sites were adjusted to avoid measuring on visual tissue boundaries. Measurement of Df_S was carried out ten times at each of the six sites (Fig. 3). Df_S was calculated as the frequency difference for t = 0 and t₁ = 30s. Time of contact between probe and sample was approximately t_c  = 2s (Fig. 1). The ten repeated measurements of each site were aimed at the same spot every time. The deviation from that spot was estimated to be less than 2 mm.   After completion of the measurements with the tactile sensor, a morphometric investigation was performed.  Tissue blocks centered round each site were embedded in paraffin.  The blocks were cut in f micron thick sections (from the upper down to the bottom surface of the measured slice) so that the tissue composition at each measuring site could be analysed.  The sections were stained with hematoxylin-eosin and studied by light microscopy.  Using a square lattice in the eyepiece of the microscope the volume density of prostate stroma, prostate glands and prostate concrements was calculated by counting the number of square-lattice intersections falling on each issue compartment, described in an earlier report (Weibel, 1979).

2.5 Statistics

Values are expressed as mean and ± SEM (standard error of mean). A one-way ANOVA procedure was used to test for differences between groups.   Pearson's correlation coefficient was used for correlation analysis. A lack of fit test was used to determine the appropriateness of linear models. Comparison of regression equations for equal slopes was made with a single multiple regression model using dummy variables and a partial F-test (Kleinbaum et al., 1998). Normality was tested using the Shapiro-Wilt test: p < 0.05 was considered statistically significant.

3. Results

3.1 Silicone model

Linear regression showed a linear relationship between Df_CTS and l_p for each muxture of silicone (R² = 0.92 – 0.97, N = 30) (Table 2 and Fig. 4).  The estimated slopes of the linear equations (Table 2) were determined not to be dependent on the mixing ratio (p = 0.61, N = 120), except for mix 4:2.5 (p = 0.0014, N = 150). There was a statistically significant dependence between the hardness of the samples and Df_CTS for all three l_ps (p = < 0.01, N = 50).

f_CTS for the unloaded condition and for conditions when the element was applied to a silicone sample were significantly different (Figs. 4 and 5); Df_CTS = -5.6^o (p < 0.001, N = 432), determined by a linear regression method with binary case variable. Thus, the frequency for which the total phase shift is zero, the zero-crossing frequency, shifted towards a lower frequency when the sensor was applied to the sample (Fig. 5).

Df_S was linearly related to the hardness of the silicone in two phases (Fig. 6). The cone standard uses a two-part cone that, for the first 150 penetration units, has a 30^o cone angle, and for penetrations higher than 150 units a 90^o cone angle.  The values for the cone penetrations exceeding 150 (Table 1) were recalculated, based on a simple cone that had a 30^o cone angle for all penetrations.   The conversion of the penetration values was performed under the assumption that the displaced volume remained constant. When Df_S was plotted against recalculated penetration values the assumption of a linear relationship was not rejected (p = 0.49, N = 50) and there was a high degree of explanation (R² = 0.98) (Fig. 6).

3.2 Prostate model

There was a large variation in Df_S between different measurement sites (Fig. 7, Table 3). The morphometric investigation of the tissue samples showed (Table 3) that there was a difference in proportions of prostate gland tissue and stroma between the different measurement sites (Fig. 8a, b). There were also variations in the amount of prostate stones.

The following equation describing the hardness of the prostate tissue based on the relative amounts of glandular tissue (Gland_%) and prostate stones (Stone_%) was proposed and used:

TissuMix = Gland% + D · Stone%

(6)

A tissue mix value in accordance with eqn. 6 was calculated for each measurement site (Fig. 7). Coefficient D was estimated in order to obtain maximum negative correlation with frequency shift. The best correlation between Df_S and the tissue mix values was found with D = -34 (R = -0.96, p < 0.001, N = 60).

a

b

Fig. 8 Sections taken from measuring sites 1(a) and 2(b) as described in Table 3 (200 X magnification). At site 1 large glands (arrows) are observed, but at site 2 there are only few glands and here prostate stones (arrows marked with s) are also detected

4. Discussion

Earlier reports have shown that Df_S and hardness are correlated in silicone models and in an animal tissue model (Lindahl and Omata, 1995). This was also confirmed in this study.  In this study it has further been demonstrated in silicone model experiments that Df_S is related to the l_p of the sensor. Under constant force application of the sensor to a sample, the basic measurement method is similar to that used for the cone penetration standard (DIN ISO 2137), where the penetration depth of a standard cone inder constant weight is the reference value. Good correlation between the international standard (manufacturers specification) and he Df_S was found.

In a recent study a tactile sensor was used to show that measurement of the stiffness of resected lymph nodes is an accurate approach to diagnosing lymph node metastases (Miyagi et al., 1997). In this study we have contributed further to the work on human tissue hardness since out in vitro prostate results indicate that relative hardness, Df_S, was correlated with the morphometrically proposed relative tissue hardness.

4.1 Silicone model

It is reasonable to assume that since the sensor tip was applied to the sample with constant force (Fig. 6), l_p must depend on the hardness of the sample.  It has been shown that for silicone samples (l_p < r₀) l_p is linearly related to Df_CTS (Fig. 4). Assuming that the resonance condition (Floyd, 1988) transfers the phase shift Df_CTS to a frequency shift Df_S (Fig. 5), we suggest that l_p, and thereby the hardness of the silicone, can be detected as a frequency shift of the sensor system.  The finidng that the sensor system could reproducibly differentiate hardness variations between silicone samples of different mixtures has previously been demonstrated (Lindahl and Omata, 1995).  In this study we went a step further with the silicone model and showed the linear relationship between Df_S and recalculated penetration values (Fig. 6), which strengthens the hypothesis that Df_S measures hardness.

As can be seen by comparing Figs. 5 and 6 there was a difference between the predicted value Df_S » 700 and the measured Df_S = 525. This is explained by the viscoelastic behaviour of silicone and the difference in time of contact, which was more than 200 s in the f_CTS measurements and approximately 28 s in the Df_S measurements.

An important parameter for the interpretation of Df_S is the contact area, S, between the sensor and the sample. Eqns. 4 and 5 show that the radius of the contact area affects what physical properties of the sample will dominate. Furthermore. The contact area is the link between the CTS and the sample.  It is reasonable to suggest that the larger the area the more of the vibrational energy will be transferred to the sample, and its influence on the CTS frequency characteristics will increase. The one-dimensionally vibrating rod theory simply accounts for one definite closing acoustic impedance and cannot explain this type of area dependence.  If the sample completely adhered to the spherical sensor tip the relationship between l_p and contact S would be linear according to:

S = 2pr0lp

(7)

There may be deviations from the relationship (eqn. 7) since it is likely that the harder silicone samples bend the surface down as a membrane rather than making a spherical cavity. The slope a of the sample with 4:2.5 mix, which was the hardest one, differed from the others (Table 2). This could be explained by the contact area differences. Also, the softer the sample the more it tends to stick to the probe resulting in increase area and possibly increased energy transfer. This could explain the dependence between sample hardness and the Df_CTS (Fig. 4) and the variation of the b–constant in Table 2. These systematic differences are not seen in the comparison between tactile sensor frequency shift and cone penetration in Fig. 6. This indicates that the same type of deviation is present in the cone penetration test of silicone.

4.2 Prostate model

With the experience gained from the silicone model experiments the experimental set-up was further developed for an in vitro human prostate model.  The measurements made with the tactile sensor on the prostate slice surface showed that there were large variations in the frequency shift between measurement sites and small differences within sites. There was, however, local variation, and it was very important to measure at the same spot when the measurements were repeated. From the silicone measurements results it was clear that with constant F_c , Df_S was related to l_p and the hardness of sample. Could this also be expected for the prostate tissue measurement? The absolute values of frequency shift from the silicone and prostate samples are not directly comparable.  This is because the force with which the sensor was applied to the sample was 8.9 times higher in the prostate set-up. In addition, the silicone samples were isotropic with acoustic impedance that is likely to vary little with mixing ration (hardness). The prostate tissue varied in composition from position to position. According to the silicone model discussion, variation in acoustic impedance affects the frequency reading even if hardness remains constant.  However, results from a study with ultrasound indicate that the relative acoustic impedance varies less than 5% in the prostate and the variation is not influenced by the presence of cancer or BPH (Tobocman et al., 1997, Fig. 5). In fact, this is one reason why it is often difficult to use ultrasound to detect differences in tissue composition in the prostate (Tobocman et al., 1997). Taking these points into account it is likely that the relative changes in frequency shift between different measurement sites reflected relative differences in hardness according to the cone penetration standard.  Since prostate, like silicone, is viscoelastic, and since we detected similar differences in Df_S in the prostate measurements as we did in the silicone measurements, it is reasonable to suggest that the hardness of prostate tissue and perhaps the hardness of human tissue in general may be related to the cone penetration standard through the regression curve of Fig. 6. This would benefit the clinical definition and understanding of tissue hardness. Of course the curve must be calibrated for the F_C of the set-up and the acoustic impedance of the measured object. The fact that the prostate tissue is non-homogenous must be further evaluated.

Morphometric analysis of the measurement sites showed that the prostate tissue was mainly composed of prostate gland tissue and stroma. It was assumed that the relative mixture of tissue components would be related to the hardness. Since glands (principally tissue fluid) and stroma (principally smooth muscle cells and fibroblasts) together make up close to 100% of the tissue, this makes Stroma_% linearly dependent on Gland_% and it is enough to consider the glands in the hardness model (eqn. 6). When prostate stones are present, as in this patient, they constitute only a fraction of 1% of the tissue, but as suggested in this study, they may influence tissue hardness and are therefore included in the model equation (eqn. 6). The prostate stone coefficient D was estimated to be 34 times greater than the gland coefficient for maximum correlation with Df_S . The common property of the two variables, Df_S and tissue mix, is hardness of prostate tissue. The results indicate that prostate glands make tissue softer while stroma and prostate stones make it harder. This suggested ability of the sensor to differentiate between tissues with histologically different composition could, hopefully, be used to objectivise the palpation of prostate cancer in the clinical situation.

The obvious question for the future is whether benign prostate hyperplasia and prostate cancer examined in vivo (not in fixed tissue as in this study) will give different frequency shifts.  The present observation that the proportion of prostate gland/stroma influences the frequency shift suggests that this could be used to guide biopsies towards areas suspicious for cancer. In general fixation hardens tissue, but whether it increase or decreases the differences is not known. In coming studies this will be explored by examining non-fixed prostate tissue.

5 Conclusions

It is concluded from this study that the catheter tip tactile sensor gives information concerning the hardness of silicone according to a cone penetration standard (DIN ISO 2137). Furthermore, human prostate model experiments on one prostate slice indicated that the relative hardness measured with the tactile sensor correlated with the proposed relative hardness of prostate tissue. This finding is promising for the further development of a non-invasive tactile sensor for detecting prostate cancer. In the clinical situation it is important for a measurable quantity to have a well-defined, sensor-independent and easily interpreted unit. The DINISO 2137 standard used here might be suitable for defining hardness of human tissue. The work will continue with further hardness calibration studies in order to link Df_S from the measurement on human tissue to this international standard. Furthermore, temperature dependence of the CTS will be investigated and measurements at body temperature will be done.

Acknowledgement - The authors would like to thank Professor Sadao Omata at the Department of Electrical Engineering, Nihon University, Japan for valuable and helpful discussion, Tomas Bäcklund at the Department of Biomedical Engineering for skillful technical assistance and useful discussions, Mrs. Birgitta Ekblom at the Department of Pathology for skillful technical assistance and Dr. Lennart Nilsson at the Department of Mathematical Studies for statistical guidance.

The study was supported by the Swedish National Board for Industrial and Technical Development (grant 95-11061), AXIOM Co Ltd., the Swedish Cancer Society (project no 1760), and the Lions Research Foundation in UmeEfont face="Times New Roman">.

References

AARNINK, R. G., Beerlage, H. P., de la Rosette, J. J. M. C. H., Debruyne, F. M. J., and Wijkstra, H. (1988): 'Transrectal ultrasound of the prostate: Innovations and future applications,' J. Urol., 159, pp. 1568-1579.

FLOYD, T. L. (1988): 'Electronic devices' (Merrill Publishing Company, Columbus, OH)

HODGE, K. K., McNeal, J. E., Terris, M. K., and Starney, T. A. (1989): 'Random systematic versus directed ultrasound guided transrectal core biopsies of the prostate,' J. Urol., 142, pp. 71-74.

KLEINBAUM, D. G., Kupper, L. L., Muller, K. E., and Nizam, A. (1998): 'Applied regression analysis and other multivariate methods' (Duxbury Press, Pacific Grove, CA)

LINDAHLl, O. A., and Omata, S. (1995): 'Impression technique for the assessment of oedema – comparison with a new tactile sensor that measures physical properties of skin,' Med. Biol. Eng. Comput., 33, pp. 27-32

LINDAHLl, O. A., Omata, S., and Ängquist, K.-A. (1998): 'A tactile sensor for detection of physical properties of human skin in vivo,' J. Med. Eng. Technol., 22, pp. 147-153

MIYAGIi, K., Furuse, A., Nakajima, J., Kohno, T., Ohtsuka, T., Yagyu, K., Oka, T., and Omata, S. (1997): 'The stiffness of lymph nodes containing lung carcinoma metastases: a new diagnostic parameter measured by a tactile sensor,' Cancer, 80, pp. 1920-1925

OMATA, S., and Terunuma, Y. (1991): 'Measurement of skin elasticity using new type tactile sensor and its application to acupuncture therapy,' Proceedings of 3rd USA-China-Japan Conference on Biomechanics, August 25-29, Georgia Institute of Technology, Atlanta, Georgia, pp. 153-154

OMATA, S., and Terunuma, Y. (1992): 'New tactile sensor like the human hand and its applications,' Sensors and Actuators A, 35, pp. 9-15

OMATA, S., and Constantinou, C. E. (1995): 'Modeling of micturition characteristics based on prostatic stiffness modulation induced using hormones and adrenergic antagonist,' IEEE Trans. Biomed. Eng., 42, pp. 843-848

TOBOCMAN, W., Resnick, M. I., and Pretlow, T. G. (1997): 'Ultrasound reflection fail to reflect the histopathology of the prostate,' Prostate, 30, pp. 33-40

US Public Health Service National Institutes of Health (1993): 'What you need to know about prostate cancer,' US-DEPT-HHS-PUBL, #NIH 93-1576 rev. ed.

WATANABE, T., Omata, S., Lee, J. Z., and Constantinou, C. E. (1997): 'Comparative analysis of bladder wall compliance based on cystometry and biosensor measurements during the micturition cycle of the rat,' Neurol. Urodynam., 16, pp. 567-581

WEIBELl, E. R. (1979): 'Stereological methods,' Vol. 1 (Academic Press, London)

Author's Biography

Anders Eklund was born in UmeEfont face="Times New Roman">, Sweden, in 1965.   He graduated with an MSc in Engineering Physics from Chalmers University of Technology, Gothenburg, Sweden, in 1993.  Between 1993 and 1995 he worked as a research engineer at the Physics Department, UmeEfont face="Times New Roman"> University. He is currently working at the Department of Biomedical Engineering and Informatics at University Hospital of Northern Sweden.  He started his PhD studies in 1997, with his main research areas concerning resonance sensors for medical diagnostics.  He is also particularly interested in the hydrodynamic characteristics of the cerebrospinal fluid system.