HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Gülü, B. Search for Related Content PubMed PubMed Citation Articles by Gülü, B.

TEACHING IN THE LABORATORY

Low-cost computer-controlled current stimulator for the student laboratory

Biomedical Engineering Institute, Boaziçi University, stanbul, Turkey

Address for reprint requests and other correspondence: B. Güçlü, Biomedical Engineering Institute, Boaziçi Univ., Bebek, Istanbul 34342, Turkey (e-mail: burak.guclu{at}boun.edu.tr)

	Abstract

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

 
Electrical stimulation of nerve and muscle tissues is frequently used for teaching core concepts in physiology. It is usually expensive to provide every student group in the laboratory with an individual stimulator. This article presents the design and application of a low-cost [about $100 (U.S.)] isolated stimulator that can be controlled by two analog-output channels (e.g., output channels of a data-acquisition card or onboard audio channels) of a computer. The device is based on a voltage-to-current converter circuit and can produce accurate monopolar and bipolar current pulses, pulse trains, arbitrary current waveforms, and a trigger output. The compliance of the current source is ±15 V, and the maximum available current is ±1.5 mA. The device was electrically tested by using the audio output of a personal computer. In this condition, the device had a dynamic range of 46 dB and the available pulse-width range was 0.1–10 ms. The device is easily programmable, and a freeware MATLAB script is posted on the World Wide Web. The practical use of the device was demonstrated by electrically stimulating the sciatic nerve of a frog and recording compound action potentials. The newly designed current stimulator is a flexible and effective tool for teaching in the physiology laboratory, and it can increase the efficiency of learning by maximizing performance-to-cost ratio.

Key words: voltage-to-current converter; audio output; neurophysiology; frog; sciatic nerve

	Introduction

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

 
BASIC PHYSIOLOGY LABORATORY EXPERIMENTS on nerve and muscle tissues are an integral part of most life science programs and biomedical engineering (1–3, 6–8). An electrical stimulator is indispensable for such experiments. Since high-end multichannel stimulators used in research can be very expensive, manufacturers also include student-grade stimulators in their product portfolios. Typical examples are the 6002 Basic Stimulator (about $1,400)¹ by Harvard Apparatus, SD9 Square-Pulse Stimulator (about $1,700) by Grass Technologies, model 2100 Isolated Pulse Stimulator ($1,800) by A-M Systems, and BSL Stimulator (about $600 but requires the BSL Basic System for about $3,000) by Biopac Systems. Although the average price of a student laboratory stimulator is lower compared with a high-end unit (e.g., DS8000 by World Precision Instruments for about $6,000), a considerable investment is required for working in small groups to achieve learning efficiency. For example, 10 units may be used in a typical class size of 20. Furthermore, the student-grade units do not always provide stimulus isolation and current output, or extra plug-in devices need to be purchased to provide those functions. They also often lack the flexibility of computer-controlled stimulus-waveform generation.

This article presents the design and application of a very low-cost (about $100) isolated current stimulator that is controlled by a personal computer. The device processes two analog voltage signals generated by a computer. The voltage signals may be obtained from a data-acquisition card (at much extra cost, e.g., for about $500–1,000 from National Instruments) or simply from the audio output that exists on all personal computers. There are also some inexpensive data-acquisition systems commercially available (e.g., for about $100–200 from Measurement Computing), but the low-cost models are generally not suitable for the stimulator device presented here, because they have low update rates for analog outputs. One of the voltage signals controls the triggering circuit of the stimulator, and the other voltage signal is scaled and converted to a current output. Since the onboard sound card of a personal computer is accessible by all high-level programming platforms, it is simple to program the current stimulator for arbitrary waveforms. A MATLAB script for pulse stimulation is provided on the World Wide Web for free access at http://web.syr.edu/bguclu/projects/stimulator. Universities typically purchase MATLAB site licenses (e.g., 10 workstation licenses for about $800) and provide this flexible computing software for use in homework and projects assigned in many different courses as well as in laboratory exercises. As an alternative, SCILAB (http://www.scilab.org) is free on the World Wide Web and offers a scientific computing platform similar to MATLAB, but it does not include as many toolboxes for specific applications. The control software for the presented device can also be written in a high-level programming language (Basic, Pascal, C, etc.) supported by personal computers. The device performs most of the functions available in the commercial units and is limited mainly by the analog output signal from the computer. There is an alternative low-cost device in the literature (4), but it requires more components and generates isolated voltage, but not current, output. Current control is important to achieve stable stimulation parameters for the physiological preparation over time.

	MATERIALS AND METHODS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

 
Electronics Design
The electronics design consists of three circuits: the isolated power supply (Fig. 1), the trigger output for external devices (Fig. 2), and the isolated current source (Fig. 3). All circuits use low-power components to minimize energy consumption. The nominal current requirement of the device is 80 mA, and the device can be powered by a commercial 9 V/100 mA mains-to-direct current (DC) adapter (not shown), which is plugged to the coaxial socket (U1) in Fig. 1. Note that the tip of the socket receives the positive terminal and the ring receives the negative terminal of the adapter. For the prototype stimulator, the circuits were built on a perforated copper board and housed in a 12 x 8-cm plastic box (Fig. 4 A). In the experiments presented here, the stimulator was controlled by the audio output of a notebook personal computer. A special cable was constructed with a stereo audio plug that split into two male BNC connectors at the other end. The parts list of the device is shown in Table 1. The pin descriptions of the integrated circuits used in the device are shown in Table 2.

View larger version (8K): [in this window] [in a new window]   Fig. 1. Schematic of the isolated power supply circuit.

View larger version (10K): [in this window] [in a new window]   Fig. 2. Schematic of the trigger output circuit.

View larger version (6K): [in this window] [in a new window]   Fig. 3. Schematic of the isolated current source circuit.

View larger version (66K): [in this window] [in a new window]   Fig. 4. Hardware and software of the current stimulator. A: the construction of the unit is simple, and the circuitry can be fitted in a plastic box. A cable carries the trigger and analog control signals from the audio jack of a notebook personal computer to two BNC connectors on the left side of the unit. The current output from the unit is received from two banana sockets on the right. The trigger output is obtained via a female BNC connector, also on the right. The light emitting diodes on the front side signal power-on and trigger output. The mains-direct current (DC) adapter is plugged to the DC power socket on the bottom side of the unit. B: the graphical user interface of StimX, a MATLAB script that can be used to generate pulse waveforms for a typical physiology experiment (see text). This MATLAB program is freely available on the World Wide Web at http://web.syr.edu/bguclu/projects/stimulator.

Trigger output.
The trigger output circuit is shown in Fig. 2. One analog output from the personal computer (e.g., the left audio channel at U2) is fed to the noninverting input of the comparator LTC1440 (IC4). The potentiometer (P1) is preset to achieve a threshold level for switching the output of the IC4 [e.g., can be adjusted to 0.2 V at the negative input voltage (V_in–)]. The output of the IC4 is +5 V (on) if the analog output from the computer is higher than the threshold and 0 V (off) if otherwise. Thus, IC4 can be controlled by software and used to trigger external devices. Each trigger output also lights the LED2 for visual feedback. R3 and R4 induce hysteresis during the on/off switching, and their selected values yield a 50-mV hysteresis voltage band. Hysteresis is used to ensure that switching is not affected by electrical noise. This is achieved by setting the voltage for off-to-on switching different (i.e., 50-mV higher) than on-to-off switching. Therefore, electrical noise cannot switch the circuit on/off unintentionally. The trigger circuit in Fig. 2 is isolated from the mains adapter ground (Fig. 1) but shares the same ground as the audio output of the personal computer.

Isolated current source.
The isolated current source is shown in Fig. 3. The second analog output from the personal computer (e.g., the right audio channel at U4) is fed to the input of unity gain isolation amplifier ISO124 (IC5). Therefore, the remaining circuitry at the output of IC5 is isolated from the personal computer as well as the external power source. The operational amplifier MC33171 (IC6) forms a second-order low-pass Butterworth filter at a 10-kHz cutoff frequency to reduce the high-frequency noise induced on the analog control signal by IC5. To have a cutoff frequency of 20 kHz, C8 can be changed to 4.7 nF and C9 to 2.2 nF. Similarly, for a cutoff frequency of 47 kHz, C8 can be changed to 2.2 nF and C9 to 1 nF. The voltage output of the filter is converted to current by instrumentation amplifier INA121 (IC7), which is used as a Howland current source. Note that the reference voltage (V_ref) output of IC7 and the current-output positive terminal (I₊) are held at the same voltage level by the second operational amplifier MC33171 (IC8). Since the first and eighth pins of IC7 are left unconnected, IC7 has unity gain, and the voltage input/output relationship can be given as follows: V_out – V_ref = V_in+ – V_in–, where V_out is the output voltage and V_in+ is the positive input voltage. Note that V_in+ is the analog control signal and V_in– is zero with respect to the isolated stimulator ground (downward arrow). Therefore, the current across the 1% tolerance resistor (R8) is V_in+ divided by the 10-k resistance. This current only flows across R8 and across the current output terminals [I₊ and current-output negative terminal (I_–)] to the isolated ground, because the input bias and offset currents of IC8 are negligible. In other words, the voltage-to-current formula is I = V/10,000, where I is the current output at I₊ and I_– and V is the voltage output at U4. For example, 0.1 V at the audio output of the personal computer generates 10-µA current. Note that the load, i.e., the impedance across the current-output terminals, does not affect the stimulator current, because the current is solely set by the control signal and R8. Therefore, the physiological preparation does not affect the performance of the presented stimulator; and, in contrast to voltage-output devices, the stimulation is precisely controlled and stable over time.

Software
A MATLAB (version 6.5, The MathWorks, Natick, MA) script, called StimX, was written to control the stimulator described above with the analog voltages at the audio output of a personal computer. The graphical user interface of the program is shown in Fig. 4B. After the program is launched with MATLAB software, calibration needs to be performed. For this purpose, a 1%-tolerance, 10-k resistor is connected to the current-output terminals (I₊ and I_–) of the stimulator, and the audio volume of the personal computer is set to a predefined level (i.e., maximum preferable). First, a high-level square wave is output from the right audio channel of the personal computer. The zero-to-peak voltage of this signal is measured on an oscilloscope and entered to the program. Then, a low-level square-wave signal is given, and the user again enters the zero-to-peak voltage. Because the audio-output characteristics of personal computers may vary, this calibration procedure ensures platform-independent operation of the stimulator. After calibration, StimX can provide a single pulse, twin pulses, or a train of current pulses by selecting the corresponding buttons in the user interface. The pulses can be positive monopolar, negative monopolar, or bipolar (the positive phase preceding the negative phase). The amplitude and duration of the pulses can be entered as decimals. If twin pulses are selected, the interval between the pulses can be specified. Additionally, if a pulse train is selected, the frequency of the pulses can be set. The adjustable delay period determines when the pulses will be started after the trigger output. In the latest version of the StimX, the trigger control signal is a 1-ms positive voltage pulse at the left audio channel of the personal computer. After stimulation settings are changed, the ‘Set’ button needs to be pressed in the StimX program for the changes to take effect. The current stimulation waveform in memory can be tested by pressing the ‘Waveform’ button. The stimulation is initiated by pressing the ‘Start’ button and terminated by pressing the same toggle button in the StimX program.

Electrical Experiments
The stimulator was first tested electrically using a 10-k dummy load resistor, and the voltages across this constant load were measured with a digital oscilloscope (model DS5102C, Rigol Technologies, Beijing, China). Although the stimulator output a controlled current through the dummy load, it was easier to find the voltage across the load and then calculate the current, because an oscilloscope measures voltage with respect to ground. The device was initially controlled by the 'daqfcngen’ function in MATLAB and used with the audio volume of the personal computer set at maximum. The frequency response of the device was measured by applying software-controlled sine waves at varying frequencies with 1-V peak amplitudes, which produced current outputs of 185 µA (peak). The linearity was tested by applying 500-Hz sine waves at different amplitudes, and the dynamic range was determined. The output of the device was also measured at different pulse stimulation parameters. Screen shots of the digital oscilloscope were captured by ScopeKit for DS5000 software (Rigol Technologies, Beijing, China).

Physiological Experiments
The practical use of the device was tested in a routine student laboratory exercise on the sciatic nerve of a frog (3). Student experiments were approved by the Institutional Ethics Committee for Animal Experiments and Care of Boaziçi University. A common water frog (Rana ridibunda) was pithed and dissected to expose the right sciatic nerve. A long segment of the sciatic nerve was severed and placed in a nerve chamber. The nerve was periodically flushed with frog's Ringer solution (112 mM NaCl, 1.9 mM KCl, 1.1 mM CaCl₂, 1.1 mM glucose, 2.4 mM NaHCO₃, and 1.0 mM NaH₂PO₄ adjusted to pH 7.2) to supply nutrients and to maintain a close-to-normal physiological condition. The nerve was stimulated at one end by the isolated current stimulator controlled by the StimX program described above. Compound action potentials (CAPs) were measured at the other end by a differential amplifier (5). CAPs were monitored on the oscilloscope and recorded on the personal computer. Specifically, the stimulus strength-duration curve was obtained, and the rheobase and chronaxie values were found.

	RESULTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

 
Frequency-Response Measurements
The frequency response of the current stimulator, which was connected to the audio channels of a personal computer, is shown in Fig. 5A; the x-axis is the frequency of the sine wave applied to the relevant audio channel of the personal computer, and the y-axis is the voltage gain (in dB units) across the dummy load. The response was approximately flat in the range of 20–5,000 Hz but attenuated at both frequency ends. This is because the audio output of a personal computer typically generates signals only in the hearing range: 20–20,000 Hz. The attenuation observed at frequencies >5,000 Hz was also due to the low-pass filter used in the isolated current source circuit (Fig. 3). It is interesting to note that the audio output applied 5-dB gain (approximately x2) to the software amplitude setting. This was probably particular to the personal computer used and may not be present with different computer hardware. The flat frequency range given above corresponds to a time constant range of 8–0.03 ms, respectively. These time constant values limited the pulse durations that could be accommodated by the current stimulator when it was controlled by the audio output of a personal computer (see below).

View larger version (10K): [in this window] [in a new window]   Fig. 5. Electrical characteristics of the current stimulator, which was connected to the audio output of a personal computer. The measurements were taken as voltages across a 10-k dummy load resistor. A: frequency response of the unit was obtained by 1-V (peak) amplitude sine waves set by software. The dotted horizontal line depicts unity gain. B: the input-output function of the unit was obtained by applying 500-Hz sine waves set by software. The dotted diagonal line depicts perfect linearity (slope: 1) with unity gain (y-intercept: 0). Vpeak, zero-to-peak voltage.

Pulse-Stimulus Measurements
The results of the pulse-stimulus measurements are given in Figs. 6 and 7. The units below the plots represent the voltage and time scales for five ticks on the oscilloscope screen shot. The current output from the stimulator was equal to the voltage across the dummy load divided by its resistance (10 k). Figure 6A shows the 5-V trigger pulse with 1-ms duration, which was also used to trigger the oscilloscope. The trigger pulse had a fast rise/fall time (<2 µs) but also some overshoot (0.4 V). At the stimulator output, a 100-µA positive single current pulse with 1-ms duration generated the voltage waveform shown in Fig. 6B across the dummy load. The rise/fall time of the current pulse was 30 µs (see also Fig. 7C). There was 7.5% maximum overshoot and undershoot in the current pulse. Additionally, because the audio output of the personal computer could not give constant-voltage output, the current pulse slightly decayed during the 1-ms on-time. The time constant for this decay was 8 ms (see also Fig. 7B). Figure 6C shows the voltage across the dummy load when a 100-µA single negative current pulse with 1-ms duration was applied, and Fig. 6D shows the bipolar version of the single pulse. Note that the duration setting in StimX refers to the duration of the positive or negative phase of a symmetric bipolar pulse (1 ms in Fig. 6D). Twin single pulses with 1-ms interpulse spacing are shown in Fig. 6E, and a 50-Hz pulse train with 2-ms positive pulses is shown in Fig. 6F.

View larger version (16K): [in this window] [in a new window]   Fig. 6. Results from the pulse stimulation experiments. The units below the plots represent the voltage and time scales for 5 ticks on the oscilloscope screen shot. The current output from the stimulator was equal to the voltage across the dummy load divided by its resistance (10 k). A: a 5-V, 1-ms trigger output pulse. B: a 100-µA positive single current pulse with 1-ms duration. C: a 100-µA negative single current pulse with 1-ms duration. D: a 100-µA bipolar single current pulse with 1-ms positive/negative phase duration. E: a 100-µA positive twin current pulses with 1-ms duration and 1-ms interpulse spacing. F: a 100-µA, 50-Hz positive pulse train with 2-ms pulse duration.

View larger version (14K): [in this window] [in a new window]   Fig. 7. More results from pulse stimulation experiments. The units below the plots represent the voltage and time scales for 5 ticks on the oscilloscope screen shot. The current output from the stimulator was equal to the voltage across the dummy load divided by its resistance (10 k). A: a 100-µA, 0.5-ms delayed single positive current pulse with 2-ms duration following the trigger pulse. B: a 100-µA single positive current pulse with 10-ms duration. C: a 100-µA single positive current pulse with 0.1-ms duration. D: a 2-µA single positive current pulse with 1-ms duration.

Physiological Experiments
The current stimulator described in this article was tested in a common physiological preparation, i.e., the sciatic nerve of a frog. The CAP generated when the nerve was stimulated with a 10-µA single positive current pulse with 1-ms duration is shown in Fig. 8A. By selecting a threshold response criterion, the stimulus amplitude and duration were covaried to obtain the stimulus strength-duration curve (Fig. 8B). The curve had features similar to those of typical electrically excitable tissues. For example, the stimulus amplitude and duration had an inverse relationship above the rheobase asymptote. That is to say, decreasing the stimulus duration required an increase in the amplitude to obtain the criterion response. The rheobase was 4.0 µA, and it was reached at 1-ms duration. Regardless of the stimulus duration, no response was obtained with stimulus amplitudes less than the rheobase. The chronaxie (i.e., the duration at twice the rheobase) was 2.8 ms.

View larger version (11K): [in this window] [in a new window]   Fig. 8. Results from the physiological experiments on a frog's sciatic nerve. A: the compound action potential (top trace: CH1) generated by a 10-µA single positive current pulse with 1-ms duration is plotted with the stimulus voltage (bottom trace: CH2). Although the stimulus current is an accurately controlled pulse, the voltage at the stimulator output (CH2) is slightly distorted, because the impedance of the nerve trunk is not purely resistive. B: stimulus strength-duration curve obtained by a threshold response criterion. The rheobase (dashed horizontal line) is 4.0 µA; the chronaxie (dashed vertical line) is 2.8 ms.

	DISCUSSION

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

Frequency Characteristics
Since the audio output of a personal computer cannot generate very-low-frequency and constant signals, current pulses with long durations could not be obtained. The practical upper limit for pulse duration was 10 ms. The performance at higher frequencies was better from the physiological point of view. Very short pulses could be obtained, and the practical lower limit was 0.1 ms. It is theoretically possible to achieve even faster response by increasing the cutoff frequency of the low-pass filter and by selecting faster electrical components. However, faster integrated circuits are more expensive, and they require more energy. On the other hand, it is more difficult to design circuits by faster discrete components (e.g., transistors). The limitation due to the sampling rate of the digital-to-analog converter, whether it is in the audio channels of the personal computer or in a stand-alone unit, is not very important in today's technology. Many commercial sound cards have sampling rates of 96 kHz and over.

Current Output
The stimulator device offers accurate current control, unlike some of the commercial units, which are based on voltage-controlled designs. Therefore, changes in the load impedance do not affect the device. Voltage-output units suffer from this problem. A limitation of the current stimulator described here, however, is its low compliance (±15 V). This means that to attain 200-µA maximum current by using the audio channel, the load resistance should be lower than 75 k. If the resistance is higher, that stimulus level cannot be obtained. The experiments presented here were limited by the amplitude of the control signal generated in the audio channel. If a stand-alone digital-to-analog card is used, ±15-V compliance can provide maximum 1.5-mA bipolar current at load resistances of <10 k. There is a simple way to increase the compliance of the stimulator by connecting multiple DC-DC converters (Fig. 1) in series. However, the maximum supply voltage rating for INA121 is ±18 V; therefore, a high-voltage instrumentation amplifier should be used in the electronics design if higher compliance is aimed. However, such amplifiers are usually not available in single commercial packages; they may be custom built from operational amplifiers.

The linearity of the current stimulator was excellent, in the range of 1–200 µA (dynamic range: 46 dB). This range is typically adequate for physiological applications (note that the rheobase was 4.0 µA in this study). The lower end of the range may be improved by filtering the background noise or using electrical components with less noise at higher cost. The resolution of the digital-to-analog converter (e.g., 16 bits) was not a limiting factor in this study. The upper end of the range may be improved by using transistors at the output stage of the stimulator, but most of the in vitro preparations do not require higher currents.

Other Physiological Preparations
The stimulator was tested thoroughly on a frog sciatic nerve preparation. However, it is a versatile device and may be used in other experiments that require electrical excitation. For example, the stimulator was qualitatively tested on the gastrocnemius muscle preparation of a frog (3). Single-twitch and tetanus responses were observed (not shown). Therefore, the current stimulator described in this article can be reliably used on many in vitro nerve and muscle preparations. However, for preparations that use high-impedance electrodes (e.g., in patch clamping), the compliance of the unit will not be adequate. As a rule of thumb, the electrode impedance should be kept at <100 k for efficient use of the device.

	GRANTS

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

 
This work was supported by Boaziçi University Research Fund 04HX101 and Turkish Scientific and Technical Research Council Grant 104S228 (to B. Güçlü).

	Acknowledgments

 
The author thanks Korcan Uçar for help with the physiological experiments and for comments on the text.

	Footnotes

 
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

¹ All currency values presented are in United States dollars.

Received for publication November 6, 2006. Accepted for publication April 11, 2007.

	REFERENCES

TOP Abstract Introduction MATERIALS AND METHODS RESULTS DISCUSSION GRANTS REFERENCES

 

1. Biopac Systems, Incorporated.PRO Lessons (online). http://www.biopac.com/Education.asp [16 April 2003].
2. D'Amour FE, Blood FR, Belden DA Jr.Manual for Laboratory Work in Mammalian Physiology (3rd ed.). Chicago, IL: Univ. of Chicago Press, 1965.
3. Güçlü B.Biomedical Instrumentation Laboratory Manuals. Istanbul: Boaziçi Univ. Press, 2006.
4. Land BR, Johnson BR, Wyttenbach RA, Hoy RR. Tools for physiology labs: inexpensive equipment for physiological stimulation. J Undergrad Neurosci Educ 3: A30–A35, 2004.
5. Land BR, Wyttenbach RA, Johnson BR. Tools for physiology labs: an inexpensive high-performance amplifier and electrode for extracellular recording. J Neurosci Methods 106: 47–55, 2001.[CrossRef][ISI][Medline]
6. Olivo RF. An online lab manual for neurophysiology. J Undergraduate Neurosci Educ 2: A16–A22, 2003.
7. Paul CA, Beltz B, Berger-Sweeney J.Discovering Neurons: the Experimental Basis of Neuroscience. Plainview, NY: Cold Spring Harbor Laboratory Press, 1997.
8. Wyttenbach R, Johnson BR, Hoy RR.Crawdad: a CR-ROM Lab Manual for Neurophysiology. Sunderland, MA: Sinauer, 1999.

This Article Abstract Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Gülü, B. Search for Related Content PubMed PubMed Citation Articles by Gülü, B.