Conditioned by the existing MRI technologies and by the available magnets, generally with a field strength of 1.5 T, in most human fMRI studies performed so far spatial resolution has ranged between 3 and 5 mm in plane, and between 3 and 10 mm along the z-axis . Functional temporal series with higher spatial resolution are compatible with fMRI experimental designs, but generally at the cost of partially sacrificing temporal resolution or SNR. On the other hand, higher spatial resolution has been achieved by many groups in human and animal studies [6-14]. Recently, even a submilli-
metre resolution in space has been reached without dramatically impairing resolution in time and SNR .
Nevertheless, further overwhelming difficulties exist, beyond the concerns about the SNR worsening, in the attempt to improve the spatial resolution. In fact, the main limitation in the spatial resolution of fMRI is not actually dependent on technical limitations but resides in the very nature of the fMRI phenomena. Indeed, since the fMRI signal arises from vascular phenomena requiring local modifications in the vascular tone, it maybe expected that spatial resolution had as a physical limit the minimal dimension of the vascular territory able to vary its blood flow independently from the neighbouring regions. This consideration has led to the hypothesis that the ultimate spatial resolution of fMRI coincides with the vascular territory of the pre-capillary arterioles, the smallest afferent vascular structures having a distinct muscular wall and so being capable of autonomous vasoregulation. Hence, the smallest functional unit visible in fMRI has been proposed to have a volume in the order of 1 mm3, and this supposed to represent the theoretical limit in spatial resolution of functional vascular phenomena .
Nevertheless, alternative hypotheses propose that an amount of capillary autoregulation is also possible through the function of pre-capillary sphincters or through an autonomous contractile function of the capillary wall. Either way, the result is a vasoregulation carried out by both the „capillary recruitment", regulating the number of capillaries open to blood flow, and by a variation in capillary diameter. While the former operates in a discrete way, in a binary „yes or no" fashion at the level of the single capillary, the latter is probably capable of a continuous regulation of the blood flow operating in a „more or less" fashion . In either case, the effect on blood flow seen above the capillary level possibly consists of a continuous variation depending on the local vascular tone.
The ancillary role of the capillary bed in regulating functional brain haemodynamics seems to be confirmed by optical spectroscopy, which produces functional maps of working mammal cortex with a precision higher than 100 |im. In these maps, functional effects have been demonstrated far beyond the resolution (1 mm3) possibly compatible with the hypothesis of the vasoregulation being operated only at the arteriolar level .
Additional evidence for this discussion arises from the analysis of the temporal dynamics of optical measurements and of blood oxygenation level dependent (BOLD) fMRI signal. While the optical spectroscopy is performed on the exposed cortex of experimental animals, the near infrared spectroscopy can be applied to humans and allows the evaluation of haemoglobin oxygenation dynamics through the intact skull. This latter technique has documented a transient rise in deoxy-
Fig. 10.1 a-e. Increased fMRI spatial specificity using the initial dip (reprinted from  with permission). In a the anatomical image of cat area 18 on the lateral gyrus (imagesize2x2 cm) is depicted. The green box indicates the field of view of the activation maps displayed in this study. b The biphasic fMRI signal time course following 10sof visual stimulation (marked in grey). c Pattern of increased positive (conventional) BOLD activity in response to moving 45° gratings. d Functional map for which only negative BOLD percentage changes occurring within the first 2 s after stimulus onset were used. e The pattern of positive BOLD responses after raising the threshold to match the number of activated pixels to that in d. It is easy to appreciate the improvement in spatial resolution in the negative BOLD functional map
Fig. 10.1 a-e. Increased fMRI spatial specificity using the initial dip (reprinted from  with permission). In a the anatomical image of cat area 18 on the lateral gyrus (imagesize2x2 cm) is depicted. The green box indicates the field of view of the activation maps displayed in this study. b The biphasic fMRI signal time course following 10sof visual stimulation (marked in grey). c Pattern of increased positive (conventional) BOLD activity in response to moving 45° gratings. d Functional map for which only negative BOLD percentage changes occurring within the first 2 s after stimulus onset were used. e The pattern of positive BOLD responses after raising the threshold to match the number of activated pixels to that in d. It is easy to appreciate the improvement in spatial resolution in the negative BOLD functional map genated haemoglobin (Hb) accompanied by a rise in oxy-Hb, which suggests an early increase in oxygen metabolism, possibly associated with an early increase in vascular volume, taking place at the capillary level, before a substantial rise in blood flow.
These data converge with the information coming from time-resolved BOLD fMRI experiments in indicating the existence of BOLD phenomena dwelling at the capillary level. Indeed, if the BOLD dynamics following a short stimulation is analysed, one can often observe an early signal deflection known as the „initial dip" phenomenon, which has recently stimulated a large body of research. The interest in the initial dip has been mainly determined by its presumed origin. Without questions, if the hypotheses concerning the dip hold true, then the phenomenon is generated by an early increase in oxygen consumption without a corresponding increase in blood flow. The occurrence of the binomial of an early metabolic increase uncoupled by a resting level regional cerebral blood flow (rCBF) is of particular interest since it is intrinsically well localized in space. In fact, an isolated increase in regional cerebral glucose metabolism (rCMRglu) is not affected by the spreading effect due to the increase in regional flow, which tends to distribute over a much more extended territory the phenomena originally circumscribed to a small cortical region.
Consequently, the use of the initial dip of the haemo-dynamic response, instead of the large later positive BOLD signal for the statistical evaluation of functional phenomena, has increased dramatically the spatial specificity of BOLD fMRI. In fact, the combined use of very high in-plane resolution (156 ^m) and of the initial dip as input to the statistical analysis, has allowed researchers to demonstrate neural activation at the columnar level in the visual cortex (V2) of anaesthetized cats at very high magnetic field strength (9.4 T) [10,11]. Taking into account the linear worsening of SNR with the reduction of voxel volume, spatial resolutions in the order of hundreds of microns are only achievable through the use of MRI units with high magnetic field intensities. The requisite of segmenting temporally the BOLD response to increasing specificity and spatial resolution clearly poses even more serious problems for the investigation of the neurophysiology of neuronal columns, since it requires keeping the temporal resolution high enough to
Fig. 10.2 a-c. fMRI-based composite angle maps of cat area 18 on the lateral gyrus (reprinted from  with permission). a The composite angle map obtained through pixel-by-pixel vector addition ofthe four single iso-orientation maps based on negative signal changes. The colour key next to a was used for colour coding the resulting orientation preferences. Overall continuity of the orientation preferences is interrupted at the orientation pinwheels where the cortical columns for different orientations are circularly arranged. The white and black circles in a depict clockwise and counterclockwise pinwheels, respectively. Two such pin-wheels are enlarged on the right (a). Scale bar for the enlarged pinwheels 200 |im. As a control, the composite maps based on MR signals obtainedbefore stimulus onset (b) and during positive BOLD signals (c) are also displayed. Maps in b and c were obtained from the same cortical region as in a. The control maps are devoid oftopological structures characteristic ofgenuine composite angle maps (A anterior, P posterior, M medial, L lateral). Scale bars (a, c) 1 mm observe and separate in time the different phases of the BOLD dynamics. This introduces a strict interdependence between spatial and temporal resolution.
The importance of achieving higher spatial resolution with higher field strength may extend beyond functional localization. In fact, the correlations between the time courses of multiple adjacent voxels in very high spatial resolution fMRI signals may contain valuable physiological information, which can be used to produce maps of neuronal-like connectivity . By grouping neighbouring voxels in relation to the degree of temporal cross-correlation between their time courses, it is possible to generate a vector field for each voxel that, for very high spatial resolution data, may present the voxel average local neuronal connectivity within its vicinity. In a rat study , it has been shown that while the amplitude changes upon stimulation of these vectors, reflecting a strengthening of local neuronal connectivity, changes in their orientation may represent the altered internal neuronal communication. Again this potential is strictly related to the capability of high-field fMRI of reducing the mixing of multiple tissue-specific signal sources for each voxel.
Was this article helpful?