- Email: [email protected]
A mobile olfactometer for fMRI-studies
Journal of Neuroscience Methods 209 (2012) 189–194
Contents lists available at SciVerse ScienceDirect
Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth
Clinical Neuroscience
A mobile olfactometer for fMRI-studies J. Ulrich Sommer a,∗ , Wakunyambo Maboshe b , Martin Griebe c , Clemens Heiser a,d , Karl Hörmann a , Boris A. Stuck a , Thomas Hummel b a
Department of Otorhinolaryngology Head and Neck Surgery, University Hospital Mannheim, Germany Smell and Taste Clinic, Department of Otorhinolaryngology, University of Dresden Medical School, Dresden, Germany Department of Neurology, University Hospital Mannheim, Germany d Department of Otorhinolaryngology, Technische Universität Munich, Munich, Germany b c
h i g h l i g h t s The olfactometer is easy to rebuild, to use and to service, is inexpensive and durable. We were able to create reproducible, fMRI measureable stimulation, using fluids. The device is fully fMRI compliant.
a r t i c l e
i n f o
Article history: Received 20 January 2012 Received in revised form 20 May 2012 Accepted 22 May 2012 Keywords: Olfactometer fMRI Self made
a b s t r a c t To perform functional magnetic resonance imaging (fMRI) studies with olfactory stimulation, the stimulation device requires special properties including those of being non-conductive and non-magnetic. It should also be easily portable and should be small enough to be stored easily when not in use. However, presently only a limited number of devices fulfill these criteria; additionally, they are typically associated with high costs. The aim of the study was to investigate whether a newly developed and relatively simple and inexpensive stimulation device would be suitable for fMRI measurements. Our stimulation device was made of standard industrial and laboratory components, has open-source software and consists of 3 core compartments namely: ‘the air inlet, control and distribution section’, ‘the odorant-section’, and ‘the delivery-section’. The device was tested in an fMRI study using 21 healthy normosmic subjects who were stimulated with two odors, d-limonene and terpinen-4-ol. Results from this trial suggest that the stimulation device is capable of creating adequate stimulation suitable for fMRI sequences. In general we describe how all sections of the olfactometer are optimized for the needs of fMRI studies. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Over the last decade, olfactory research has encountered an increased surge and caliber of exploration. This has catered to major advancements in the field including the regular use of fMRI in olfactometry and the construction of commercially available stimulation devices (so-called “olfactometers”). Functional magnetic resonance imaging (fMRI) has allowed for the characterization of brain structures that are related to olfaction by identifying patterns of brain activation after the administration of olfactory stimuli. The basis behind the idea to build a mobile olfactometer stemmed from the fact that despite the fore mentioned advancements, a need for inexpensive and easy to handle fMRI olfactometers still remains.
∗ Corresponding author at: Department of Otorhinolaryngology Head and Neck Surgery, University Hospital Mannheim, Theodor-Kutzer-Ufer 1-3, 68167 Mannheim, Germany. Tel.: +49 0 621 383 1600; fax: +49 0 621 383 3827. E-mail address: [email protected] (J.U. Sommer). 0165-0270/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jneumeth.2012.05.026
We founded our stimulation device on the following criteria: inexpensive (about 1700 USD); easily serviceable; safe and durable; MRI compliant; easily storable; as well as being simple to use and to calibrate. This manuscript describes a stimulation device that meets the demands of the above criteria (Fig. 1). With the provided part list it is possible to reproduce the olfactometer with basic technical skills. We describe the basic functional and structural elements of the mobile stimulation device and present data proofing the functional performance of the device.
2. Materials and methods 2.1. Olfactometer design and construction 2.1.1. Airflow The stimulation device depicted here is molded around the general construct of most olfactometers i.e. it can be divided into 3 subsections: the “air inlet, control and distribution section”, the
190
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
Fig. 1. Front view of the complete stimulation device.
“odorant-section” and the “delivery-section”. Airflow is at the very essence of stimulus delivery in olfactometry. As described by Kobal (1981) or Johnson et al. (2006), in olfactometers, the carrier gas is directed through an odorant canister where it becomes odorized. This odorized airflow is diluted with a clean airflow and is thereafter delivered to the subjects’ nose as a temperature controlled, humidified odorant stimulus. For some MRI studies, the airflow may not be thermostabilized, nor have a high flow rate (Lorig et al., 1999) as is so with chemosensory event related potentials (CSERP) (Kobal and Hummel, 1988). A high temporal resolution is not crucial in fMRI studies due to the relatively slow nature of the hemodynamic response and thus a more simple stimulus delivery unit is sufficient to get fMRI results. It is for this reason that our device features a pure and simple approach to the airflow apparatus. In our design, air is taken from a regular clean air wall outlet and via computer controlled electro-pneumatic valves is directed to 4 gas-washing bottles. These bottles contain the desired liquid odorants and the control fluids as shown in Fig. 2. 2.1.1.1. Air inlet and flow-control. As air enters the olfactometer, a constant airflow is achieved using a ball-flow-meter. However, it is important to note that this airflow-controlling unit is not part of the olfactometer. Using a standard wall-mounted air-outlet (providing odorless, clean, dry air), found in most infirmaries, a flow
meter can be plugged in. The flow meter we used consisted of a glass tube containing a ball that floated on the stream of air (AIR LIQUIDE Deutschland GmbH, D-40235 Duesseldorf). Using such a device means that the stream of air entering the olfactometer is adjustable between 2 and 15 l/m. Following entry into the olfactometer, the air is directed into a solenoid operated three-way pneumatic switching valve, the socalled “Airflow-valve” Fig. 2(1). In its turned-off-state, the air leaves the valve through the normally connected (NC) port. The NC port is not associated with a hose, so it functions as an exhaust and maintains the steady airflow. Thus, a pressure control unit is not required, only the airflow needs to be regulated. When the Airflow-valve is turned on, the air leaves this valve passing the normally open (NO) port, which is split into four hoses. This separates the air into four streams and distributes it to four solenoid operated three-way pneumatic switching valves, the so called “Odorant valves” also depicted in Fig. 2(2). This design is inspired from the previous work of Lorig et al. but changes the used setup in a crucial point so that the airflow switching valves are put before the odorant chambers (Lorig et al., 1999). This way the valve has not to be exchanged when the used odorant is changed. Thus, the approach is radically different from the one used by (Kobal, 1981), as solenoid valves and no vacuum lines are used for switching of the airflow.
Fig. 2. Schematic diagram showcasing the general layout of the pneumatic system inside the olfactometer including the fluid/odor reservoirs, the valve system and their connection to the air inlet (seen at the base of the diagram). NC = normally connected port; NO = normally open port; 1 = air-flow valve; 2 = odorant valves; 3 = fluid/odorant bottles.
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
2.1.1.2. Odorant section. In this section of the device, the air is saturated with odorant as it travels through the porous frit at the bottom of the gas-washing bottles and forms small air-bubbles. The odorant valves are directly connected to four gas-washing bottles (Standard taper joint 29/32, 100 ml – neoLab, Heidelberg, Germany) filled with liquid odorants. These bottles are equipped with frit cartridges (Standard taper joint 29/32, Por 1 – neoLab, Heidelberg, Germany) with the air bubbling through the frits such that the air is exposed to a maximally large surface of liquid odorant. A short piece of silicone tubing fits neatly in the frit cartridges, so no further elements are necessary to connect the PTFE tubing. All the parts in contact with the odorants are standard lab equipment, thus making them easily accessible and exchangeable. 2.1.1.3. Delivery section. Upon exiting the 4 gas washing chambers, odorized air or control air, respectively, passes through 4 individual PTFE lines to the delivery section. After 10 cm – just before entry into the subject’s nose, the gas-flows are connected (using t-fittings) to a single PTFE-flow line with a length of 5 cm (see pneumatics section below). For stimulus presentation all odorants pass through the same line in the final 5 cm of the tubing so that the dead space is kept to a minimum. At the end of the tubing a nosepiece for monorhinal stimulation, a mask or a Y-piece can be attached, to deliver odorants to one or both nostrils, according to the individual experiment setup. The olfactometer uses a continuous airflow design and as a result, regardless of how many valves are simultaneously opened, the resulting airflow at the delivery unit always matches the one entering the device avoiding air-puffs at the outlet. 2.1.2. Hardware and pneumatic system Portability is an essential feature for an imaging stimulation device. Our stimulation device is small and compact enough for easy transportation (to and from MRI facilities) and storage. The whole olfactometer was built in a standard 19 aluminum case (GEH SG 119, reichelt elektronik GmbH & Co. KG, Sande, Germany) to ensure a rugged and yet mobile design and is only 42 cm × cm 11 cm × 23 cm (W × H × D) in size with a weight of approximately 3 kg. Fig. 3B: depicts an overhead view of the main interior of the stimulation device when the top cover is opened. On top of the casing, 4 holes, each of 48 mm in diameter were cut into the aluminum cover. The edges of the holes were covered with regular edge protectors. The pneumatics unit was attached to an aluminum bracket, fixed to the lower cross brace of the casing as shown above in Fig. 3A.
191
Apart from the air-inlet portion, all tubing within the olfactometer, including air lines and odor lines are made using standard PTFE tubing with 3 mm bore and 4 mm outer diameter (neoLab, Heidelberg, Germany). The PTFE tube also acts as an attachment for the nosepiece and is the final tube before entry into the subject’s nose. The hoses leave and enter the olfactometer through brass pushin system connectors (M5, 2-6290, neoLab, Heidelberg, Germany). The valves are 24 V, solenoid operated, monostable, ‘Food and Drug Administration’ (FDA) conform, 3/2 way pneumatic valves (ETV-117C30 – AVS-Römer GmbH, Koenigsdorf, Germany), which completely isolate the airflow from the solenoid and the switching mechanism. 2.1.2.1. Electric components. The olfactometer relies on 3-way pneumatic valves to regulate its airflow. These electromechanical devices are electrically controlled using a computer. The power supply and the USBc interface board were built on eurocards (100 mm × 160 mm) and located in standard 19 subrack modules. A 110-230VAC-24VDC open frame switching power supply (PSA075 – HN Electronic Components GmbH, Langenselbold, Germany) was used and because of the inductive load of the solenoids, an accessory film capacitor (680nF [C1]) and a surge protection diode (1.5KE 27A [D6]) were utilized. The USB–computer interface was affiliated with an IO-Warrior24 module (Code Mercenaries, Schönefeld, Germany) used in a ready-made relay circuit board (RK10-05USBV1 – GWR-Elektronik, Völklingen, Germany). All internal wiring was prepared with 1.0 mm2 strand and the internal connections were made with Faston type blade-connectors. The complete circuit diagram of the device is shown in Fig. 4 A ‘Manual Airflow Control Panel’ was installed to monitor the status of the solenoid valves with LEDs and also to activate them independent of a connected computer by means of rocker-switches. The classic 5 mm LEDs were attached to the solenoid’s positive terminal with appropriate current limiting resistors [R1–R5] (1.2 k for blue and 1.5 k for red, yellow and green). 2.1.3. Software The control software provides a means for the user to program the desired settings e.g. flow rate, stimulus duration and valve status. The software used (including the controlling program’s source-code) is available online for free. Coding was done in GNU C++ and to be freely available, the source code was released under the GNU General Public License (GPL) on http://github.com/sommeru/riech-o-mat. The attached files
Fig. 3. (A) The pneumatics section of the mobile olfactometer. (B) Overhead view of the olfactometer. With the top cover opened, the odorant section can be viewed. OR = odorant reservoirs.
192
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
Fig. 4. Schematic of the olfactometer electric circuits.
include the source code, the binary, the so called ‘make files’ and additional files such as icons. The program can be compiled under most POSIX conform operating systems, including Linux (Kernel 2.6+). Using IO-Warriors libraries (Code Mercenaries/Schönefeld/Germany/www.codemercs.com), the controlling computer can access the solenoid valves with standard Universal Serial Bus (USB) connectors. The control program allows the user to define the valve status using a 5 digit binary code. An exemplary code would be “11000”. The first number sets the “Airflow valve” to ON and the last 4 numerals are setting “Odor valve” 1 to ON and the “Odor valves” 2–4 to OFF. This binary code can be programmed and customized according to the aims and needs of its user. After a certain “stepping time”, the next 5-bit-code is read from a file and the next valve positions are set. Acceptable stepping times range from 50 ms to 10 s. 2.2. Olfactometer setup To set up the olfactometer, the gas-washing bottles should be filled with roughly 50 ml of liquid odorant or control fluid and installed into the system. The long PTFE hoses must then be attached to both the gas-washing bottles and t-fittings. The air-inlet can now be connected to a wall mounted clean air source or a scuba diving tank and the flow rate should be set to about 2–3 l/min. All “Odor-valves” should be opened and the ‘Airflow-Rocker-Switch’ is set to ON. Afterwards, the olfactometer is ready for use.
replaced when exchanging the odorant in the corresponding gaswashing bottle. In addition, none of the airflow controlling parts must be replaced each time odorants are exchanged. 2.4. Olfactometer characteristics The olfactometer characteristics were recorded using a hotfilm anemometer as a mass flow meter (Honeywell AWM-Series AWM5102 VN, Honeywell International, Morristown, NJ, USA). With a precision of ±0.5%, airflow could be recorded at 0–10 l/min and a response of <60 ms. The device delivers trapezoid shaped airflow pulses with rise and fall times of about 120 ms after valve switching. This delay seems to be stable, as long as no change of inlet-airflow is made. The stability of the stimulus is only affected by the quality and constancy of the supplied air. Using a standard ball flow meter, the airflow is constant at 2 ± 0.1 l/min over 1 h. Differences of the airflow between control and odor bottles were below 6% and not noticeable by the subjects if the airflow was turned off before switching. Because of major drops in airflow (>12%) switching of the odorants without “off-phase” is not recommended. This may be due to air compression effects in the odor chambers and tubing. Due to the trapezoid shaped airflow pulses and the major drops in airflow when directly switching channels the olfactometer cannot be used for recording of CSERPs. 3. Experimental setup 3.1. Initial fMRI testing
2.3. Olfactometer maintenance The olfactometer offers an advantage of being low maintenance (Johnson et al., 2006). Each time the odorants are exchanged, new gas-washing bottles should be used. Moreover, only one PTFE tube must be used for one odorant. Accordingly, it should also be
3.1.1. Subjects Twenty-one healthy, normosmic volunteers participated. The recruiting and scanning was performed at the Department of Otorhinolaryngology, Smell and Taste Clinic, University of Dresden Medical School, Dresden, Germany. In an fMRI session, scanning
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
193
Fig. 5. Block 1 with ‘stimulation’ and ‘control’ part used for the fMRI sessions. The experimental setup consisted of a block design with 6 ‘stimulation’ and with 6 ‘control’ blocks.
took place while stimulating the subjects alternatingly with terpinenol and limonene from our stimulation device. The results reported here are parts of a larger study the results of which are reported elsewhere (Sezille et al., 2012). 3.1.2. Olfactometer As previously mentioned, the odorants were delivered to each subject by our custom-built fMRI stimulation device. A small foamplug was attached to the end of the PTFE delivery tubes and thus acted as the nosepiece in these sessions. Using a Y-piece odors were presented to both nostrils. The total flow rate was set to 2 l/min. Besides 50 ml of odorant, 50 ml of distilled water was used as the control fluid. Therefore only two of four gas-washing bottles were used. 3.1.3. fMRI Blood-oxygen level dependent (BOLD) fMRI was performed using a 1.5 Tesla scanner (Siemens Sonata) in a block design consisting of 6 stimulation blocks, each lasting 42 s. Each block consisted of 21 s with and 21 s without stimulus. For registration, 88 volumes per session with a 26 axial-slice matrix 2D SE/EP sequence (matrix: 64 × 64; TR: 3000 ms; TE: 45ms; FA: 90◦ ; voxel size: 3 mm × 3 mm × 3 mm; FOV: 192), were used. By turning the airflow off for 2 s after 1 s of odor/control airstream, the switch between ‘stimulation’ and the ‘control’ blocks became undetectable by the subjects. Due to the low airflow (1 l/min per nostril) the airstream did not cause significant trigeminal stimulation; specifically, none of the participants reported nasal congestion, secretion or irritation after the measurements. The sensations possibly caused by the very slight airflow changes were present in both, ‘stimulation’ and ‘control’ conditions, and therefore did not affect final analyses. Fig. 5 illustrates the design of block 1 of 6. The succeeding 5 other blocks were constructed in the same manner. The statistical evaluation of the data of 21 subjects was performed by the MATLAB based SPM8 software package. Statistical threshold was 0.005 (10 voxels, uncorrected). fMRI data were superimposed on a T1 anatomical reference.
4.1. Activation with terpinenol and limonene: initial trial Brain activations in human subjects in response to the odorants terpinenol and limonene in orbitofrontal cortex (OFC), piriform cortex (PIR) and cerebellum (CER) are shown in Fig. 6. Areas of significant activation were identified at cluster level for values exceeding a p-value of 0.005 (10 voxels, uncorrected). The uncorrected p was chosen to be able to show typical activation to olfactory stimuli in several typical brain areas. These activations are in line with previous studies on olfactory activation that associate these brain areas with olfactory sensations (Gottfried, 2006; Wang et al., 2012). The findings suggest that it is possible to self-construct a serviceable stimulation device from regular laboratory components and deliver useful results. As previously expressed, our main aim was to build a mobile olfactometer that was MRI amenable. Our stimulation device offers aforementioned advantages such as easy operation. On the other hand, the mobile olfactometer clearly has limitations. For instance, the device is only compatible with liquid or solid odorants and does not allow the delivery of gaseous odorants. In addition, it is only possible to deliver fractions of dilutions of the used odorants (100%, 50%, 33%, 25%) by opening multiple bottles of control fluid (H2 O) at once. Finally, it may only be operated with low airflows. Accordingly, the system cannot be used within a setup
4. Results and discussion The need for inexpensive, commercially available olfactometers is one that is acknowledged in the field of olfactometry. Several groups including the recent publication by Lundström et al. (2010) have put forward templates for interested researchers to selfconstruct their own stimulation devices and overcome the existing financial barriers. On a parallel level, this manuscript offers the basic outline for such a cause; conversely, our device is of a much simpler design in order to target its use in MRI research and not to evaluate CSERPs.
Fig. 6. Averaged brain activations in human subjects in response to the odorants terpinenol and limonene in orbitofrontal cortex (OFC), piriform cortex (PIR) and cerebellum (CER). Areas of significant activation were identified at cluster level for values exceeding a p-value of 0.005 (10 voxels, uncorrected). The uncorrected p was chosen to be able to show typical activation to olfactory stimuli in several typical brain areas. fMRI data were collected on a 1.5 T Siemens Sonata scanner in 88 volumes/session with a 26 axial-slice matrix 2D SE/EP sequence (matrix: 64 × 64; TR: 3000 ms; TE: 45ms; FA: 90◦ ; voxel size: 3 mm × 3 mm × 3 mm; FOV: 192).
194
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
for olfactory ERP; it may also be problematic in experiments based on an event-related MRI design. 5. Conclusions Typical brain responses to olfactory stimuli were observed using fMRI. Results from the trial indicated that our easy-to-use, easyto-build and cost-effective stimulation device is able to create reproducible and adequate stimulation with liquid odorants. In summary, all parts of the ‘mobile olfactometer’ are designed to be inexpensive, easily serviceable, durable and MRI-compliant. We hope that this manuscript can deliver to those who wish to have a simple MRI-olfactometer to be used in block designs. Acknowledgments We are indebted to Caroline Sezille and Moustafa Bensafi in analyzing results from fMRI measurements. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jneumeth.2012.05.026.
References Gottfried JA. Smell: central nervous processing. Adv Otorhinolaryngol 2006;63:44–69. Johnson BN, Russell C, Khan RM, Sobel N. A comparison of methods for sniff measurement concurrent with olfactory tasks in humans. Chem Senses 2006;31:795–806. Kobal G. Elektrophysiologische untersuchungen des menschlichen geruchssinns. Stuttgart: Thieme; 1981. Kobal G, Hummel C. Cerebral chemosensory evoked potentials elicited by chemical stimulation of the human olfactory and respiratory nasal mucosa. Electroencephalogr Clin Neurophysiol 1988;71:241–50. Lorig TS, Elmes DG, Zald DH, Pardo JV. A computer-controlled olfactometer for fMRI and electrophysiological studies of olfaction. Behav Res Methods Instrum Comput 1999;31(2):370–5. Lundström JN, Gordon AR, Alden EC, Boesveldt S, Albrecht J. Methods for building an inexpensive computer-controlled olfactometer for temporally-precise experiments. Int J Psychophysiol 2010;78(2): 179–89. Sezille C, Chakirian A, Thevenet M, Gerber J, Hummel T, Bensafi M. Complexity of odorant structure influences human olfactory cortex activity: an fMRI study. In: Abstract presented at the annual meeting of the association for chemoreception sciences; 2012. Wang J, Eslinger PJ, Doty RL, Zimmerman EK, Grunfeld R, Sun X, et al. Olfactory deficit detected by fMRI in early Alzheimer’s disease. Brain Res 2012;1357: 184–94.
Contents lists available at SciVerse ScienceDirect
Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth
Clinical Neuroscience
A mobile olfactometer for fMRI-studies J. Ulrich Sommer a,∗ , Wakunyambo Maboshe b , Martin Griebe c , Clemens Heiser a,d , Karl Hörmann a , Boris A. Stuck a , Thomas Hummel b a
Department of Otorhinolaryngology Head and Neck Surgery, University Hospital Mannheim, Germany Smell and Taste Clinic, Department of Otorhinolaryngology, University of Dresden Medical School, Dresden, Germany Department of Neurology, University Hospital Mannheim, Germany d Department of Otorhinolaryngology, Technische Universität Munich, Munich, Germany b c
h i g h l i g h t s The olfactometer is easy to rebuild, to use and to service, is inexpensive and durable. We were able to create reproducible, fMRI measureable stimulation, using fluids. The device is fully fMRI compliant.
a r t i c l e
i n f o
Article history: Received 20 January 2012 Received in revised form 20 May 2012 Accepted 22 May 2012 Keywords: Olfactometer fMRI Self made
a b s t r a c t To perform functional magnetic resonance imaging (fMRI) studies with olfactory stimulation, the stimulation device requires special properties including those of being non-conductive and non-magnetic. It should also be easily portable and should be small enough to be stored easily when not in use. However, presently only a limited number of devices fulfill these criteria; additionally, they are typically associated with high costs. The aim of the study was to investigate whether a newly developed and relatively simple and inexpensive stimulation device would be suitable for fMRI measurements. Our stimulation device was made of standard industrial and laboratory components, has open-source software and consists of 3 core compartments namely: ‘the air inlet, control and distribution section’, ‘the odorant-section’, and ‘the delivery-section’. The device was tested in an fMRI study using 21 healthy normosmic subjects who were stimulated with two odors, d-limonene and terpinen-4-ol. Results from this trial suggest that the stimulation device is capable of creating adequate stimulation suitable for fMRI sequences. In general we describe how all sections of the olfactometer are optimized for the needs of fMRI studies. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Over the last decade, olfactory research has encountered an increased surge and caliber of exploration. This has catered to major advancements in the field including the regular use of fMRI in olfactometry and the construction of commercially available stimulation devices (so-called “olfactometers”). Functional magnetic resonance imaging (fMRI) has allowed for the characterization of brain structures that are related to olfaction by identifying patterns of brain activation after the administration of olfactory stimuli. The basis behind the idea to build a mobile olfactometer stemmed from the fact that despite the fore mentioned advancements, a need for inexpensive and easy to handle fMRI olfactometers still remains.
∗ Corresponding author at: Department of Otorhinolaryngology Head and Neck Surgery, University Hospital Mannheim, Theodor-Kutzer-Ufer 1-3, 68167 Mannheim, Germany. Tel.: +49 0 621 383 1600; fax: +49 0 621 383 3827. E-mail address: [email protected] (J.U. Sommer). 0165-0270/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jneumeth.2012.05.026
We founded our stimulation device on the following criteria: inexpensive (about 1700 USD); easily serviceable; safe and durable; MRI compliant; easily storable; as well as being simple to use and to calibrate. This manuscript describes a stimulation device that meets the demands of the above criteria (Fig. 1). With the provided part list it is possible to reproduce the olfactometer with basic technical skills. We describe the basic functional and structural elements of the mobile stimulation device and present data proofing the functional performance of the device.
2. Materials and methods 2.1. Olfactometer design and construction 2.1.1. Airflow The stimulation device depicted here is molded around the general construct of most olfactometers i.e. it can be divided into 3 subsections: the “air inlet, control and distribution section”, the
190
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
Fig. 1. Front view of the complete stimulation device.
“odorant-section” and the “delivery-section”. Airflow is at the very essence of stimulus delivery in olfactometry. As described by Kobal (1981) or Johnson et al. (2006), in olfactometers, the carrier gas is directed through an odorant canister where it becomes odorized. This odorized airflow is diluted with a clean airflow and is thereafter delivered to the subjects’ nose as a temperature controlled, humidified odorant stimulus. For some MRI studies, the airflow may not be thermostabilized, nor have a high flow rate (Lorig et al., 1999) as is so with chemosensory event related potentials (CSERP) (Kobal and Hummel, 1988). A high temporal resolution is not crucial in fMRI studies due to the relatively slow nature of the hemodynamic response and thus a more simple stimulus delivery unit is sufficient to get fMRI results. It is for this reason that our device features a pure and simple approach to the airflow apparatus. In our design, air is taken from a regular clean air wall outlet and via computer controlled electro-pneumatic valves is directed to 4 gas-washing bottles. These bottles contain the desired liquid odorants and the control fluids as shown in Fig. 2. 2.1.1.1. Air inlet and flow-control. As air enters the olfactometer, a constant airflow is achieved using a ball-flow-meter. However, it is important to note that this airflow-controlling unit is not part of the olfactometer. Using a standard wall-mounted air-outlet (providing odorless, clean, dry air), found in most infirmaries, a flow
meter can be plugged in. The flow meter we used consisted of a glass tube containing a ball that floated on the stream of air (AIR LIQUIDE Deutschland GmbH, D-40235 Duesseldorf). Using such a device means that the stream of air entering the olfactometer is adjustable between 2 and 15 l/m. Following entry into the olfactometer, the air is directed into a solenoid operated three-way pneumatic switching valve, the socalled “Airflow-valve” Fig. 2(1). In its turned-off-state, the air leaves the valve through the normally connected (NC) port. The NC port is not associated with a hose, so it functions as an exhaust and maintains the steady airflow. Thus, a pressure control unit is not required, only the airflow needs to be regulated. When the Airflow-valve is turned on, the air leaves this valve passing the normally open (NO) port, which is split into four hoses. This separates the air into four streams and distributes it to four solenoid operated three-way pneumatic switching valves, the so called “Odorant valves” also depicted in Fig. 2(2). This design is inspired from the previous work of Lorig et al. but changes the used setup in a crucial point so that the airflow switching valves are put before the odorant chambers (Lorig et al., 1999). This way the valve has not to be exchanged when the used odorant is changed. Thus, the approach is radically different from the one used by (Kobal, 1981), as solenoid valves and no vacuum lines are used for switching of the airflow.
Fig. 2. Schematic diagram showcasing the general layout of the pneumatic system inside the olfactometer including the fluid/odor reservoirs, the valve system and their connection to the air inlet (seen at the base of the diagram). NC = normally connected port; NO = normally open port; 1 = air-flow valve; 2 = odorant valves; 3 = fluid/odorant bottles.
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
2.1.1.2. Odorant section. In this section of the device, the air is saturated with odorant as it travels through the porous frit at the bottom of the gas-washing bottles and forms small air-bubbles. The odorant valves are directly connected to four gas-washing bottles (Standard taper joint 29/32, 100 ml – neoLab, Heidelberg, Germany) filled with liquid odorants. These bottles are equipped with frit cartridges (Standard taper joint 29/32, Por 1 – neoLab, Heidelberg, Germany) with the air bubbling through the frits such that the air is exposed to a maximally large surface of liquid odorant. A short piece of silicone tubing fits neatly in the frit cartridges, so no further elements are necessary to connect the PTFE tubing. All the parts in contact with the odorants are standard lab equipment, thus making them easily accessible and exchangeable. 2.1.1.3. Delivery section. Upon exiting the 4 gas washing chambers, odorized air or control air, respectively, passes through 4 individual PTFE lines to the delivery section. After 10 cm – just before entry into the subject’s nose, the gas-flows are connected (using t-fittings) to a single PTFE-flow line with a length of 5 cm (see pneumatics section below). For stimulus presentation all odorants pass through the same line in the final 5 cm of the tubing so that the dead space is kept to a minimum. At the end of the tubing a nosepiece for monorhinal stimulation, a mask or a Y-piece can be attached, to deliver odorants to one or both nostrils, according to the individual experiment setup. The olfactometer uses a continuous airflow design and as a result, regardless of how many valves are simultaneously opened, the resulting airflow at the delivery unit always matches the one entering the device avoiding air-puffs at the outlet. 2.1.2. Hardware and pneumatic system Portability is an essential feature for an imaging stimulation device. Our stimulation device is small and compact enough for easy transportation (to and from MRI facilities) and storage. The whole olfactometer was built in a standard 19 aluminum case (GEH SG 119, reichelt elektronik GmbH & Co. KG, Sande, Germany) to ensure a rugged and yet mobile design and is only 42 cm × cm 11 cm × 23 cm (W × H × D) in size with a weight of approximately 3 kg. Fig. 3B: depicts an overhead view of the main interior of the stimulation device when the top cover is opened. On top of the casing, 4 holes, each of 48 mm in diameter were cut into the aluminum cover. The edges of the holes were covered with regular edge protectors. The pneumatics unit was attached to an aluminum bracket, fixed to the lower cross brace of the casing as shown above in Fig. 3A.
191
Apart from the air-inlet portion, all tubing within the olfactometer, including air lines and odor lines are made using standard PTFE tubing with 3 mm bore and 4 mm outer diameter (neoLab, Heidelberg, Germany). The PTFE tube also acts as an attachment for the nosepiece and is the final tube before entry into the subject’s nose. The hoses leave and enter the olfactometer through brass pushin system connectors (M5, 2-6290, neoLab, Heidelberg, Germany). The valves are 24 V, solenoid operated, monostable, ‘Food and Drug Administration’ (FDA) conform, 3/2 way pneumatic valves (ETV-117C30 – AVS-Römer GmbH, Koenigsdorf, Germany), which completely isolate the airflow from the solenoid and the switching mechanism. 2.1.2.1. Electric components. The olfactometer relies on 3-way pneumatic valves to regulate its airflow. These electromechanical devices are electrically controlled using a computer. The power supply and the USBc interface board were built on eurocards (100 mm × 160 mm) and located in standard 19 subrack modules. A 110-230VAC-24VDC open frame switching power supply (PSA075 – HN Electronic Components GmbH, Langenselbold, Germany) was used and because of the inductive load of the solenoids, an accessory film capacitor (680nF [C1]) and a surge protection diode (1.5KE 27A [D6]) were utilized. The USB–computer interface was affiliated with an IO-Warrior24 module (Code Mercenaries, Schönefeld, Germany) used in a ready-made relay circuit board (RK10-05USBV1 – GWR-Elektronik, Völklingen, Germany). All internal wiring was prepared with 1.0 mm2 strand and the internal connections were made with Faston type blade-connectors. The complete circuit diagram of the device is shown in Fig. 4 A ‘Manual Airflow Control Panel’ was installed to monitor the status of the solenoid valves with LEDs and also to activate them independent of a connected computer by means of rocker-switches. The classic 5 mm LEDs were attached to the solenoid’s positive terminal with appropriate current limiting resistors [R1–R5] (1.2 k for blue and 1.5 k for red, yellow and green). 2.1.3. Software The control software provides a means for the user to program the desired settings e.g. flow rate, stimulus duration and valve status. The software used (including the controlling program’s source-code) is available online for free. Coding was done in GNU C++ and to be freely available, the source code was released under the GNU General Public License (GPL) on http://github.com/sommeru/riech-o-mat. The attached files
Fig. 3. (A) The pneumatics section of the mobile olfactometer. (B) Overhead view of the olfactometer. With the top cover opened, the odorant section can be viewed. OR = odorant reservoirs.
192
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
Fig. 4. Schematic of the olfactometer electric circuits.
include the source code, the binary, the so called ‘make files’ and additional files such as icons. The program can be compiled under most POSIX conform operating systems, including Linux (Kernel 2.6+). Using IO-Warriors libraries (Code Mercenaries/Schönefeld/Germany/www.codemercs.com), the controlling computer can access the solenoid valves with standard Universal Serial Bus (USB) connectors. The control program allows the user to define the valve status using a 5 digit binary code. An exemplary code would be “11000”. The first number sets the “Airflow valve” to ON and the last 4 numerals are setting “Odor valve” 1 to ON and the “Odor valves” 2–4 to OFF. This binary code can be programmed and customized according to the aims and needs of its user. After a certain “stepping time”, the next 5-bit-code is read from a file and the next valve positions are set. Acceptable stepping times range from 50 ms to 10 s. 2.2. Olfactometer setup To set up the olfactometer, the gas-washing bottles should be filled with roughly 50 ml of liquid odorant or control fluid and installed into the system. The long PTFE hoses must then be attached to both the gas-washing bottles and t-fittings. The air-inlet can now be connected to a wall mounted clean air source or a scuba diving tank and the flow rate should be set to about 2–3 l/min. All “Odor-valves” should be opened and the ‘Airflow-Rocker-Switch’ is set to ON. Afterwards, the olfactometer is ready for use.
replaced when exchanging the odorant in the corresponding gaswashing bottle. In addition, none of the airflow controlling parts must be replaced each time odorants are exchanged. 2.4. Olfactometer characteristics The olfactometer characteristics were recorded using a hotfilm anemometer as a mass flow meter (Honeywell AWM-Series AWM5102 VN, Honeywell International, Morristown, NJ, USA). With a precision of ±0.5%, airflow could be recorded at 0–10 l/min and a response of <60 ms. The device delivers trapezoid shaped airflow pulses with rise and fall times of about 120 ms after valve switching. This delay seems to be stable, as long as no change of inlet-airflow is made. The stability of the stimulus is only affected by the quality and constancy of the supplied air. Using a standard ball flow meter, the airflow is constant at 2 ± 0.1 l/min over 1 h. Differences of the airflow between control and odor bottles were below 6% and not noticeable by the subjects if the airflow was turned off before switching. Because of major drops in airflow (>12%) switching of the odorants without “off-phase” is not recommended. This may be due to air compression effects in the odor chambers and tubing. Due to the trapezoid shaped airflow pulses and the major drops in airflow when directly switching channels the olfactometer cannot be used for recording of CSERPs. 3. Experimental setup 3.1. Initial fMRI testing
2.3. Olfactometer maintenance The olfactometer offers an advantage of being low maintenance (Johnson et al., 2006). Each time the odorants are exchanged, new gas-washing bottles should be used. Moreover, only one PTFE tube must be used for one odorant. Accordingly, it should also be
3.1.1. Subjects Twenty-one healthy, normosmic volunteers participated. The recruiting and scanning was performed at the Department of Otorhinolaryngology, Smell and Taste Clinic, University of Dresden Medical School, Dresden, Germany. In an fMRI session, scanning
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
193
Fig. 5. Block 1 with ‘stimulation’ and ‘control’ part used for the fMRI sessions. The experimental setup consisted of a block design with 6 ‘stimulation’ and with 6 ‘control’ blocks.
took place while stimulating the subjects alternatingly with terpinenol and limonene from our stimulation device. The results reported here are parts of a larger study the results of which are reported elsewhere (Sezille et al., 2012). 3.1.2. Olfactometer As previously mentioned, the odorants were delivered to each subject by our custom-built fMRI stimulation device. A small foamplug was attached to the end of the PTFE delivery tubes and thus acted as the nosepiece in these sessions. Using a Y-piece odors were presented to both nostrils. The total flow rate was set to 2 l/min. Besides 50 ml of odorant, 50 ml of distilled water was used as the control fluid. Therefore only two of four gas-washing bottles were used. 3.1.3. fMRI Blood-oxygen level dependent (BOLD) fMRI was performed using a 1.5 Tesla scanner (Siemens Sonata) in a block design consisting of 6 stimulation blocks, each lasting 42 s. Each block consisted of 21 s with and 21 s without stimulus. For registration, 88 volumes per session with a 26 axial-slice matrix 2D SE/EP sequence (matrix: 64 × 64; TR: 3000 ms; TE: 45ms; FA: 90◦ ; voxel size: 3 mm × 3 mm × 3 mm; FOV: 192), were used. By turning the airflow off for 2 s after 1 s of odor/control airstream, the switch between ‘stimulation’ and the ‘control’ blocks became undetectable by the subjects. Due to the low airflow (1 l/min per nostril) the airstream did not cause significant trigeminal stimulation; specifically, none of the participants reported nasal congestion, secretion or irritation after the measurements. The sensations possibly caused by the very slight airflow changes were present in both, ‘stimulation’ and ‘control’ conditions, and therefore did not affect final analyses. Fig. 5 illustrates the design of block 1 of 6. The succeeding 5 other blocks were constructed in the same manner. The statistical evaluation of the data of 21 subjects was performed by the MATLAB based SPM8 software package. Statistical threshold was 0.005 (10 voxels, uncorrected). fMRI data were superimposed on a T1 anatomical reference.
4.1. Activation with terpinenol and limonene: initial trial Brain activations in human subjects in response to the odorants terpinenol and limonene in orbitofrontal cortex (OFC), piriform cortex (PIR) and cerebellum (CER) are shown in Fig. 6. Areas of significant activation were identified at cluster level for values exceeding a p-value of 0.005 (10 voxels, uncorrected). The uncorrected p was chosen to be able to show typical activation to olfactory stimuli in several typical brain areas. These activations are in line with previous studies on olfactory activation that associate these brain areas with olfactory sensations (Gottfried, 2006; Wang et al., 2012). The findings suggest that it is possible to self-construct a serviceable stimulation device from regular laboratory components and deliver useful results. As previously expressed, our main aim was to build a mobile olfactometer that was MRI amenable. Our stimulation device offers aforementioned advantages such as easy operation. On the other hand, the mobile olfactometer clearly has limitations. For instance, the device is only compatible with liquid or solid odorants and does not allow the delivery of gaseous odorants. In addition, it is only possible to deliver fractions of dilutions of the used odorants (100%, 50%, 33%, 25%) by opening multiple bottles of control fluid (H2 O) at once. Finally, it may only be operated with low airflows. Accordingly, the system cannot be used within a setup
4. Results and discussion The need for inexpensive, commercially available olfactometers is one that is acknowledged in the field of olfactometry. Several groups including the recent publication by Lundström et al. (2010) have put forward templates for interested researchers to selfconstruct their own stimulation devices and overcome the existing financial barriers. On a parallel level, this manuscript offers the basic outline for such a cause; conversely, our device is of a much simpler design in order to target its use in MRI research and not to evaluate CSERPs.
Fig. 6. Averaged brain activations in human subjects in response to the odorants terpinenol and limonene in orbitofrontal cortex (OFC), piriform cortex (PIR) and cerebellum (CER). Areas of significant activation were identified at cluster level for values exceeding a p-value of 0.005 (10 voxels, uncorrected). The uncorrected p was chosen to be able to show typical activation to olfactory stimuli in several typical brain areas. fMRI data were collected on a 1.5 T Siemens Sonata scanner in 88 volumes/session with a 26 axial-slice matrix 2D SE/EP sequence (matrix: 64 × 64; TR: 3000 ms; TE: 45ms; FA: 90◦ ; voxel size: 3 mm × 3 mm × 3 mm; FOV: 192).
194
J.U. Sommer et al. / Journal of Neuroscience Methods 209 (2012) 189–194
for olfactory ERP; it may also be problematic in experiments based on an event-related MRI design. 5. Conclusions Typical brain responses to olfactory stimuli were observed using fMRI. Results from the trial indicated that our easy-to-use, easyto-build and cost-effective stimulation device is able to create reproducible and adequate stimulation with liquid odorants. In summary, all parts of the ‘mobile olfactometer’ are designed to be inexpensive, easily serviceable, durable and MRI-compliant. We hope that this manuscript can deliver to those who wish to have a simple MRI-olfactometer to be used in block designs. Acknowledgments We are indebted to Caroline Sezille and Moustafa Bensafi in analyzing results from fMRI measurements. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jneumeth.2012.05.026.
References Gottfried JA. Smell: central nervous processing. Adv Otorhinolaryngol 2006;63:44–69. Johnson BN, Russell C, Khan RM, Sobel N. A comparison of methods for sniff measurement concurrent with olfactory tasks in humans. Chem Senses 2006;31:795–806. Kobal G. Elektrophysiologische untersuchungen des menschlichen geruchssinns. Stuttgart: Thieme; 1981. Kobal G, Hummel C. Cerebral chemosensory evoked potentials elicited by chemical stimulation of the human olfactory and respiratory nasal mucosa. Electroencephalogr Clin Neurophysiol 1988;71:241–50. Lorig TS, Elmes DG, Zald DH, Pardo JV. A computer-controlled olfactometer for fMRI and electrophysiological studies of olfaction. Behav Res Methods Instrum Comput 1999;31(2):370–5. Lundström JN, Gordon AR, Alden EC, Boesveldt S, Albrecht J. Methods for building an inexpensive computer-controlled olfactometer for temporally-precise experiments. Int J Psychophysiol 2010;78(2): 179–89. Sezille C, Chakirian A, Thevenet M, Gerber J, Hummel T, Bensafi M. Complexity of odorant structure influences human olfactory cortex activity: an fMRI study. In: Abstract presented at the annual meeting of the association for chemoreception sciences; 2012. Wang J, Eslinger PJ, Doty RL, Zimmerman EK, Grunfeld R, Sun X, et al. Olfactory deficit detected by fMRI in early Alzheimer’s disease. Brain Res 2012;1357: 184–94.