This article shows an optimized procedure for imaging of the neural substrates of auditory stimulation in the songbird brain using functional Magnetic Resonance Imaging (fMRI). It describes the preparation of the sound stimuli, the positioning of the subject and the acquisition and subsequent analysis of the fMRI data.
The neurobiology of birdsong, as a model for human speech, is a pronounced area of research in behavioral neuroscience. Whereas electrophysiology and molecular approaches allow the investigation of either different stimuli on few neurons, or one stimulus in large parts of the brain, blood oxygenation level dependent (BOLD) functional Magnetic Resonance Imaging (fMRI) allows combining both advantages, i.e. compare the neural activation induced by different stimuli in the entire brain at once. fMRI in songbirds is challenging because of the small size of their brains and because their bones and especially their skull comprise numerous air cavities, inducing important susceptibility artifacts. Gradient-echo (GE) BOLD fMRI has been successfully applied to songbirds 1-5 (for a review, see 6). These studies focused on the primary and secondary auditory brain areas, which are regions free of susceptibility artifacts. However, because processes of interest may occur beyond these regions, whole brain BOLD fMRI is required using an MRI sequence less susceptible to these artifacts. This can be achieved by using spin-echo (SE) BOLD fMRI 7,8 . In this article, we describe how to use this technique in zebra finches (Taeniopygia guttata), which are small songbirds with a bodyweight of 15-25 g extensively studied in behavioral neurosciences of birdsong. The main topic of fMRI studies on songbirds is song perception and song learning. The auditory nature of the stimuli combined with the weak BOLD sensitivity of SE (compared to GE) based fMRI sequences makes the implementation of this technique very challenging.
1. Preparation of the Auditory Stimuli
2. Subject Preparation
2.1 Subject and group size
Here we present a protocol specifically adapted to the use of (adult) zebra finches. The choice of the species depends on the scientific question. However, other considerations like bird robustness to anaesthesia may also be taken into account. Zebra finches (Taeniopygia guttata) should be housed in aviaries under a 12 hr light: 12 hr dark photoperiod and have access to food and water ad libitum throughout the study. The minimal number of individuals per experiment is 15. This number takes into account the sensitivity of spin-echo fMRI and the natural inter-individual variability of biological phenomena measured in the experiment.
2.2 Installation of setup and preparation of the animal
(For specification of the used equipment, we refer to the list of specific reagents and equipment at the end of this article)
As mentioned above, a 7T MR system is used in the presented setup. Other MR systems with different field strengths are also possible, but at 7T a good compromise is reached between signal-to-noise ratio and degree of susceptibility artifacts (see discussion). At higher field strengths the signal-to-noise ratio will increase together with the degree of susceptibility artifacts.
3. Data Acquisition
4. Data Processing
An alternative to this is the use of ‘SPMMouse’ which is a toolbox allowing SPM to open and analyze files of any voxel dimension. The tool allows SPM ‘glass brains’ to be created from any image, and automatically adjusts defaults length scales based on the headers of image files or user entered data. Hence, this toolbox works in the opposite way than what we propose. Instead of changing the voxel size of the images to fit in SPM, the default settings of SPM are changed to use images with different voxel sizes.
We here visually presented an optimized sequence of procedures for successful imaging of neural substrates of auditory stimuli in the zebra finch brain. Firstly, the described procedure for preparation of the auditory stimuli results in stimuli that can be incorporated into an ON/OFF block paradigm (Figure 2) and that are normalized to eliminate potential differences in sound pressure level that could evoke a differential response in the brain. After preparing the zebra finch for MRI scanning and positioning it into the bore of the magnet (Figure 1), fMRI can be acquired. Additionally, a 3D high-resolution image is taken in order to normalize the data to the zebra finch atlas 9. Finally, pre-processing and statistical analysis of the data allows visualization of the results obtained (Figure 6).
Figure 1. Spectrograms of white noise recorded in order to establish frequency bands that are enhanced/suppressed within the magnet bore. A. White noise outside the magnet bore. B. White noise recorded at the location of the bird’s head inside the magnet bore. C. White noise after application of the equaliser function to correct for enhanced/suppressed frequency bands.
Figure 2. Overview of the ON/OFF block paradigm in which auditory stimulation periods are alternated with rest periods. Each block (stimulus/rest) lasts 16 sec during which 2 images are acquired. The different stimuli consist of representative motifs of birdsong or other types of sound depending on the experiment. These motifs are concatenated and interleaved with silent periods and the duration of the silent periods is adjusted to keep the total amount of sound and silence identical over all stimuli.
Figure 3. Setup for auditory fMRI in small songbirds. A. Animal bed. Inset: Detailed schematic overview of the positioning of the bird in the animal bed of the scanner: B. RF head coil, C. Beak mask with D. supply of anaesthetic gas, E. non-magnetic headphones, F. pneumatic pillow sensor to monitor the respiration rate, G. cloacal temperature probe, H. feedback controlled heater system to keep body temperature of the bird stable during the measurement. Click here to view larger figure.
Figure 4. Slice geometry for whole-brain fMRI imaging. Composition of screenshots from geometry editor in ParaVision software. Previously acquired axial, sagittal and coronal RARE piloting images are used to define the slice orientation for the fMRI scan.
Figure 5. Lateral view of a 3D representation of the left hemisphere with delineated structures from the zebra finch atlas 9, projected on its mid-sagittal slice. The color code of the delineated nuclei is presented on the right. These delineated structures are part of the vocal motor pathway: HVC, nucleus robustus arcopallii (RA), nXII pars tracheosyringealis (nXIIts); the anterior forebrain pathway: nucleus lateralis magnocellularis pars lateralis (LMAN), area X (X); the auditory system: field L, nucleus ovoidalis (Ov), nucleus mesencephalicus lateralis pars dorsalis (MLd); the olfactory system: olfactory bulb (OB); and the visual system: nucleus entopalliallis (E), tectum opticum (TeO).
Figure 6. Example of an fMRI BOLD response in the primary auditory region, Field L, and adjacent secondary auditory regions evoked by different auditory stimuli compared to the rest condition. The images consist of statistical parametric maps superimposed on high-resolution anatomical images from the zebra finch brain atlas 9. T-values are colour coded according to the scale displayed in the figure and only voxels in which the t-test was found to be significant (p<0.001) are displayed.
In this report, we describe an optimized protocol for the detailed in vivo characterization of neural substrates of auditory stimulation in anaesthetized zebra finches.
In line with the presented protocol, the majority of functional brain activation studies in animals using BOLD fMRI, anaesthetize the animals during the acquisition. Training animals to accustom them to the magnet environment and the scanner noise during the study periods is also possible but rather time-consuming and challenging and therefore rarely employed.
Although anaesthesia minimizes stress-induced effects on the physiological responses of interest and facilitates animal handling, its effect both on the neural response and on the transfer function between the neural activity and the BOLD response measured in fMRI is an on-going and important research topic. Therefore, the effects of anaesthesia on the BOLD response during auditory stimulation in zebra finches were investigated in our lab 2. Accordingly, three widely used anaesthetics in zebra finches – medetomodine, isoflurane and urethane – acting on different neurotransmitter systems, were studied. The results indicated that auditory stimulation resulted in clear BOLD responses with all three anaesthetics, but that slight differences occurred between the three reagents in relation to e.g. extension of the area of activation. Based on the results of this study and on the fact that isoflurane is the most common anaesthetic in clinical applications as it has the great advantage of having relatively rapid recovery and minor side effects and thus has the highest potential for use in longitudinal studies, isoflurane became the anaesthetic of choice for zebra finch fMRI at our lab.
In this protocol we use spin-echo (SE) fMRI instead of the more traditional gradient-echo (GE) fMRI. Compared to GE fMRI, SE fMRI has the great advantage of providing signal to the whole brain as there is no signal dropout in the images. Another advantage of SE BOLD fMRI is its better spatial specificity 10,11 . Indeed, at high magnetic field, the intravascular component of the SE BOLD signal is reduced (because of a long TE) and the extravascular component from large vessels is suppressed (by the 180 ° refocusing pulse of the SE MRI sequence). The SE BOLD signal is thus dominated by an accurate extravascular signal originating from small vessels 12-14. The main limitation of SE fMRI is its relatively weak sensitivity, requiring optimized sequences and optimized stimulation paradigms. The contrast to noise ratio (CNR) increases with field strength 15. A long TE also increases the CNR, but compromises the signal-to-noise ratio 12,13,15 . The optimal TE usually corresponds to a time equal to or longer than the T2 value of the tissues. We have shown that, at 7T, a TE value of 60 msec provides a CNR and a signal-to-noise ratio sufficient to detect significant differences in BOLD responses triggered by different stimuli (Poirier, 2010).
Compared to GE T2*-weighted contrast, SE T2-weighted contrast requires a long TR (1,500-2,000 msec at 7T). To be able to image 15 slices, we used a TR of 2,000 msec. To keep the acquisition time at a reasonable limit, SE MRI sequences need to be speeded up. This is usually achieved using the echo planar imaging (EPI) sampling scheme 10,16-19 . However, EPI induces image distortions that increase with the magnitude of the magnetic field, and contaminates the BOLD signal with T2* effects (making the signal stronger but less specific). EPI also produces a very intense acoustic noise, making it less relevant for use in investigating auditory stimuli. We thus used a RARE sequence with a matrix size of 64 x 32, which resulted in an acquisition time of 8 sec. This temporal resolution is still compatible with the sluggish BOLD response induced by block designs, but too slow to accurately sample the time course of the BOLD response or to use event-related designs. With this sequence, we thus obtained a pure T2-weighted SE signal, which is characterized by a very good spatial specificity, a sensitivity sufficiently high to detect differential BOLD responses and a temporal resolution compatible with the used stimulus paradigm 20,21 .
Advantages and Limitations of the use of fMRI in Songbirds
During the last decades, fMRI has become one of the most popular neuroimaging techniques in clinical cognitive neuroscience for the study of brain activity during various tasks ranging from simple sensory-motor to highly cognitive tasks. In preclinical research, this method is, however, still only scarcely used. The scarcity of fMRI experiments completed in small animals and particularly songbirds to date possibly relates to the fact that anaesthesia or sedation is required to achieve complete immobilization of the subjects (see discussion above). Hence, this is considered the major drawback of the technique and restricts the type of questions that can be addressed. However, although fMRI requires anaesthesia and the BOLD signal mainly reflects local field potentials and thus differs from the action potentials measured in electrophysiological and Immediate early gene (IEG) studies (e.g. 22), BOLD fMRI has confirmed many results obtained by these techniques.
To date, the most popular techniques in songbird neuroscience are still activity-dependent expression of IEG and electrophysiological recordings of single- or multi-unit activity. These techniques benefit from a very high spatial resolution (5-30 μm; cellular level). However, they are highly invasive or even lethal. Additionally, electrophysiological techniques are limited by the number of locations that can be sampled in one experiment and thus require a priori hypotheses about the localization of the neuronal substrate involved in the investigated process. In contrast, BOLD fMRI allows a whole-brain approach – with a spatial resolution of 250 μm – and can thus be used to perform assumption-free experiments. Finally and most importantly, the non-invasiveness of MRI enables repeated longitudinal measures on the same subjects, which opens a large range of new possibilities.
The authors have nothing to disclose.
This research was supported by grants from the Research Foundation – Flanders (FWO, project Nr G.0420.02 and G.0443.11N), the Hercules Foundation (grant Nr AUHA0012), Concerted Research Actions (GOA funding) from the University of Antwerp, and partly sponsored by EC – FP6 project DiMI, LSHB-CT-2005-512146 and EC – FP6 project EMIL LSHC-CT-2004-503569 to A.VdL. G.DG and C.P. are Postdoctoral Fellows of the Research Foundation – Flanders (FWO).
Name of the reagent/equipment | Company | Catalogue number | Comments |
Isoflurane anaesthetic | Isoflo | 05260-05 | |
PC-Sam hardware/software | SA-Instruments | http://www.i4sa.com | |
Monitoring and gating system | 1025 | ||
MR-compatible small rodent heater system | Model 1025 compatible | ||
Rectal temperature probe | RTP-102B | 7”, 0.044” | |
7T MR scanner | Bruker Biospin | PHS 70/16 | |
Paravision software | 5.1 | ||
Gradient Insert | BGA9S | 400 mT/m, 300A, 500V | |
Gradient Amplifiers | Copley Co., USA | C256 | |
Transmit resonators | Inner diameter: 72 mm, transmit only, active decoupled | ||
Receiver antenna – 20 mm quadrature Mouse Head | Receive only, active decoupled | ||
WaveLab software | Steinberg | ||
Praat software | Paul Boersma, University of Amsterdam | http://www.praat.org | |
Non-magnetic dynamic speakers | Visation, Germany | HK 150 | |
Fiber optic microphone | Optoacoustics, | Optimic 1160 | |
Sound amplifier | Phonic corporation | MM 1002a | |
Presentation software | Neurobehavioral Systems Inc. | ||
MRIcro | Chris Rorden | http://www.cabiatl.com/mricro/mricro/ | |
Statistical Parametric Mapping (SPM) | Welcome Trust Centre for Neuroimaging | 8 | http://www.fil.ion.ucl.ac.uk/spm/ |