Research articleLaminar specificity in monkey V1 using high-resolution SE-fMRI
Introduction
Functional MRI can only be truly useful if the physiological origin of the measured signals is thoroughly understood. However, it is necessary to achieve the spatiotemporal resolution that permits visualization of functional units, such as the cortical columns or laminae, to relate the fMRI signal to the neurophysiological events underlying it.
The cortex is divided horizontally in laminae, which share similar properties and functionality [1], and vertically in columns. The cortical columns are considered the functional processing units in the cortex [2], [3], [4]. The size of the layers and columns of different species ranges from 50 μm to 1 mm. Consequently, in addition to signal-to-noise considerations, the signal's spatial specificity becomes a predominant resolution-limiting factor. Specificity depends on the vascularization density of an area, on the static magnetic field, as well as on the employed pulse sequence. The conventional gradient-echo (GE) fMRI, the point-spread function (PSF) of which is estimated to be of the order of 1.5 to 3 mm at high field [5], [6], [7], [8], [9], clearly falls short in resolving the cortical microarchitecture of the primate brain. BOLD signal is an indirect measure of neural activity that is based on the dynamics of deoxyhemoglobin (dHb), which acts as an endogenous contrast agent. Not surprisingly, therefore, the spatial resolution that can be achieved with BOLD contrast is limited by the physiological and anatomical properties of the vascular bed. Neural activity causes an increase in local blood flow, which results in a decreased concentration of dHb. This decreases the local static susceptibility gradients (characterized by T2*) and leads to a signal increase.
The GE-EPI is sensitive to these local susceptibility gradients, and at high magnetic fields, the signal originates predominantly from susceptibility effects near small draining venules and veins. Because, however, the concentration of dHb at any given site depends on the cerebral blood flow (CBF) within these venules and veins, the measured signal changes do not necessarily coincide with the location of the neural activity, but they can be also detected on sites distal to the neural activation. Convergence of draining venules from activated regions is likely to generate signal changes of even higher magnitude than those measured at the activation site. It has indeed been reported that the strongest functional signal measured with GE-EPI tends to occur in the upper layers and at the surface of the cortex because of the high density of draining veins found there [10], [11]. That the fMRI signal originates near veins can also be seen at ultra-high-resolution MRI studies, where functional activation was shown to originate from within or around small veins penetrating the cortex [12], [13].
To increase the spatial resolution of the fMRI activation maps, by tackling the brain–vein problem, several strategies have been successfully employed. They include (a) using methods based on the subtraction of functional signals elicited with opposite stimulus conditions [14], [15], [16], [17]; (b) considering in the analysis only the early part of the functional response, which is thought to be more capillary specific, and thus determined to a lesser extent by the effects of venous drainage [17], [18], [19]; (c) using the highest possible magnetic field strength which diminishes the relative contribution of larger veins; and (d) appropriate selection of pulse sequences. The current report concentrates on the application of spin-echo (SE) BOLD in the visual cortex of the monkey.
Regions surrounding the vessels are characterized by local susceptibility gradients. As mentioned previously, signal in GE sequences will reflect the strength of these gradients, with larger signal loss for more extensive susceptibility gradients. Spin-echo sequences, on the other hand, can refocus such gradients and prevent signal loss, provided that the spins in the tissues around the vessels are static or their average diffusion-induced displacement is very small compared to the local magnetic field changes. It is therefore thought that the SE-BOLD signal changes are due to the average water diffusion [20], [21], [22], [23], [24], [25] and is dependent on the apparent T2. So, in other words, if the diffusion rate of the spins is of comparable size as the fluctuations in the magnetic field caused by the inhomogeneity, the spins cannot be effectively refocused [21], [22], [23]. In addition to the diffusional effects, an intravascular effect also contributes to the SE-BOLD signal, especially at low magnetic fields. This contribution is diminished at high magnetic fields due to the short T2 and T2* of blood.
As a result of the above mentioned effects, while the GE-BOLD signal is strongest at the surface of the brain, where a larger density of veins is found [10], [26], the SE-BOLD signal is confined in the gray matter [11], [27], [28], [29]. The PSF of the latter is estimated to be 1 mm or less [6], [8], [30], [31]. On the other hand, the SE signal is smaller than the GE signal, and therefore, high magnetic fields are needed. Thus, SE-fMRI may be more suitable for high-resolution applications at high field.
Experiments from different laboratories corroborate this notion. For instance, ocular dominance columns have been shown with SE-EPI in humans at 7 T using a differential stimulus paradigm [31], [32]. In contrast, however, despite the origination of SE-BOLD in deeper cortical layers compared to the GE signal [11], [27], [28], [29], segregation of laminae with SE-BOLD fMRI was less successful.
The global aim of our project is to elucidate the origin of SE-BOLD fMRI by directly comparing signal changes in small regions, that is, possibly confined within single ocular dominance columns (ca. 400 μm) or laminae (ca. 100–500 μm), with the electrical signals obtained in extracellular recordings. Therefore, we set out first to optimize SE-BOLD in monkeys in experiments at 4.7 T. Here we report our first results with high-resolution SE-fMRI of the primary visual cortex (V1) in the macaque, aiming to visualize functional activity in different cortical layers. The strongly layered structure of V1 in the macaque is ideally suited for this. In V1, the vasculature shows marked differences in the different layers in humans [33] and monkeys [34].
A number of variables, including field strength, pulse sequence and its parameters, may yield different activation maps by representing more strongly the venous compartments or the capillaries [22], [23], [27], [35], [36], [37]. The consensus is that at high field, the GE BOLD signal arises mostly from static susceptibility effects around small veins, while the SE-BOLD is thought to be sensitive to capillaries. However, we show here that, in many cases, the SE-EPI is still sensitive to T2* effects and, as such, represents a substantial venous fraction [38], [39]. We also addressed the main disadvantage of the SE, its low temporal resolution. This was overcome by using a modified stimulus/data acquisition design that increases the effective temporal resolution, thus, allowing high-spatial and high-temporal resolution SE-fMRI.
Section snippets
Subjects
Experiments were performed in six healthy monkeys (Macaca mulatta) weighing 4.5–7 kg. All experiments were approved by local authorities (Regierungspräsidium) and were in full compliance with the guidelines of the European community (EUVD 86/609/EEC) for the care and use of laboratory animals. The anesthesia procedure and experimental setup are described in detail in Refs. [16], [40]. Briefly, animals were intubated after induction with fentanyl (3 μg/kg), thiopental (5 mg/kg) and succinyl
Spin-echo fMRI
Fig. 3 shows high-resolution anatomical and functional images of macaque striate cortex acquired with the GE fast imaging sequence (Bruker) — a customized FLASH sequence [41]. The high-resolution GE anatomical image at 100×100-μm spatial resolution clearly shows the layered structure of the primary visual cortex (Fig. 3A). Gray and white matter can be easily distinguished but, in addition the Gennari line, can be seen as a darker line. The Gennari line is confined to cortical layer IV and is
Discussion
We have shown that the spatial specificity of SE-fMRI is sufficient to show the cortical laminae for in-plane spatial resolutions in the range of 250 to 500 μm and slice thickness of 2 mm. Based on a comparison with high-resolution anatomical images, the location of the functional activation could be determined. In all animals, functional activity mainly originated from the cortical layer IV/Duvernoy layer 3. Layer IV is the input layer, where thalamic synapses terminate. It has a high density
Conclusion
Using SE-fMRI, high spatial, high temporal resolution, SE-fMRI data was obtained. Multishot SE allows for high-spatial resolution, and temporal resolution was increased by using a different stimulus paradigm. We reliably obtained high-resolution SE-fMRI data, and cortical layers could be clearly discriminated. Functional activation was predominantly observed in layer IV of the striate cortex. Further research will focus on obtaining high spatiotemporal resolution data and relating this to the
Acknowledgments
We would like to thank Mark Augath for technical support and Kathrin Guhr, Mirko Lindig and Christian Samhaber for preparing and maintaining animal anesthesia. The Max-Planck Society supported the work.
References (50)
- et al.
How accurate is magnetic resonance imaging of brain function?
Trends Neurosci
(2003) - et al.
Spatial specificity of high resolution gradient echo (GE) BOLD and spin echo (SE) BOLD fMRI in cat visual cortex at 9.4 T
- et al.
High-resolution functional magnetic resonance imaging of the animal brain
Methods
(2003) - et al.
Human ocular dominance columns as revealed by high-field functional magnetic resonance imaging
Neuron
(2001) - et al.
Ultra high-resolution fMRI in monkeys with implanted RF coils
Neuron
(2002) - et al.
Functional brain mapping by blood oxygenation level-dependent contrast magnetic resonance imaging. A comparison of signal characteristics with a biophysical model
Biophys J
(1993) - et al.
Combined imaging-histology study of cortical laminar specificity of fMRI signal
Neuroimage
(2006) - et al.
Cortical blood vessels of the human brain
Brain Res Bull
(1981) - et al.
Signal and noise characteristics of Hahn SE and GE BOLD fMRI at 7 T in humans
Neuroimage
(2005) - et al.
The critical relationship between the timing of stimulus presentation and data acquisition in blocked designs with fMRI
Neuroimage
(1999)