Abstract—Cole-Cole parameters of 40 post-menopausal women are compared with their DEXA bone mineral density measurements. Impedance characteristics of four extremities are compared; left and right extremities are statistically same, but lower extremities are statistically different than upper ones due to their different fat content. The correlation of Cole-Cole impedance parameters to bone mineral density (BMD) is observed to be higher for dominant arm. With the post-menopausal population, ANOVA tests of the dominant arm characteristic frequency, as a predictor for DEXA classified osteopenic and osteoporic population around lumbar spine, is statistically very significant. When used for total lumbar spine osteoporosis diagnosis, the area under the Receiver Operating Curve of the characteristic frequency is 0.830, suggesting that the Cole-Cole plot characteristic frequency could be a useful diagnostic parameter when integrated into standard screening methods for osteoporosis. Moreover, the characteristic frequency can be directly measured by monitoring frequency driven angular behavior of the dominant arm without performing any complex calculation. Keywords—Bio-impedance spectroscopy, bone mineral density, characteristic frequency, osteoporosis, receiver operating curve. I. INTRODUCTION UBJECTS are women age 40 and above, with no hip or arm fracture history, no metabolic bone diseases. Women with no menstruation for at least 12 months are classified as postmenopausal [1]. The experiments are conducted with 40 menopausal women, according to Declaration of Helsinki and all subjects have signed an informed consent form prior to participating in this study. To minimize overnight fasting effects on BI measurements trunk measurements are ignored; the dominant body segment for impedance measurement is right arm, for right handed subjects. Following their body impedance measurements in sitting position, subjects are scanned in the bone densitometer to obtain reference BMD and T scores from L1-L4 lumbar spine and from hip. T score is the standard deviation departure from BMD mean of the young normal population and based on World Health Organization (WHO) Study Group classifications. T scores are verified by a radiologist. The Dual-energy X-ray Absorptiometry (DEXA) machine measurement precision is 1.14% or better. For T ≥ -1 the subject is classified as “Normal”; for -2.5 < Τ < -1 F. Matur was with the Institute of Biomedical Engineering in Bogazici University. He is now with Xerox Turkey (Phone: +90 212-354-7000; e-mail: firat.matur@xerox.com). Y. Ülgen is with the Institute of Biomedical Engineering in Bogazici University (phone: +90 216-516-3432; e-mail: ulgeny@boun.edu.tr). “Osteopenic”, and for T ≤ -2.5 as “Osteoporic”. II. METHODOLOGY Body Impedance measurements are performed by using the HP4284A Precision LCR Meter, at 20 different frequencies between 10 kHz to 800 kHz; with frequencies evenly distributed over the logarithmic scale [2]. Based on EN60601 electrical safety standard, the maximum current that is allowed to pass through a patient is limited to 100μA at 1 kHz [3]. Tetra-polar measurement model is used to eliminate the electrode-skin impedances [4], [5]. To ensure measurement stability of HP4284A during measurements, internal A/D integration level set is set to medium to allow 180 ms time window. Four consecutive measurements are averaged to obtain the final reading [6]. In impedance measurements, standard Ag/AgCl Arbo Kendall electrodes (48 x 34 mm) are used and no additional lubricant is applied to electrode contact points. Electrodes are placed at the outer side of the hands along the 3rd metacarpal bones, and over the feet, along the feet 3rd metacarpal bones. Current is applied to anterior electrodes while voltages are measured from the posterior ones. By using 4 electrodes and interchanging electrode positions, it is possible to measure arm, leg or trunk impedances separately [7]. To reduce effect of overnight fasting on BI measurements, trunk impedance measurements are ignored. Although, different tissues will contribute to measured impedance proportionally to their volume share, the body segment can be modeled as a  ௘ resistance in parallel to a  ௜ serial to a  ஼௉஺ . The impedance is calculated as in (1), where  ௜ is intracellular fluid resistance,  ௘ is the extracellular fluid resistance,  ଴ and  ஶ are extrapolated resistance values obtained from Cole-Cole plot where plot crosses real axis, ߙis the depression angle constant and ܭis a constant with dimension of (ohm.sec-α),  is the angular frequency [8], [9]. The frequency at which imaginary part of impedance model is minimized, is the Cole-Cole characteristic frequency  ௖ .   ) ( 1 0 * j K R R R R R Z e i        (1) Capacitive effects of cell membranes are usually lumped in constant phase angle impedance  ஼௉஺ as defined in     ) ( j K Z CPA (2) Screening Post-Menopausal Women for Osteoporosis by Complex Impedance Measurements of the Dominant Arm Fırat Matur, Yekta Ülgen S World Academy of Science, Engineering and Technology International Journal of Biomedical and Biological Engineering Vol:9, No:9, 2015 699 International Scholarly and Scientific Research & Innovation 9(9) 2015 scholar.waset.org/1307-6892/10002362 International Science Index, Biomedical and Biological Engineering Vol:9, No:9, 2015 waset.org/Publication/10002362