Wouter Schellekens1, Martijn Thio1, Nick Ramsey2, and Natalia Petridou1
1Radiology, UMC Utrecht, Utrecht, Netherlands, 2Neurosurgery, UMC Utrecht, Utrecht, Netherlands
Synopsis
In the current study, we apply population Receptive
Field modeling to somatosensory vibrotactile stimulation, using 7 Tesla
functional MRI. We find somatotopic structures within primary somatosensory
cortical areas BA3, BA1, and BA2. Furthermore, the receptive field sizes describe
tactile information integration and allows for the direct assessment of
processing hierarchy within primary somatosensory cortex. This finding is
further supported by the estimated HRFs, showing that the BOLD response is
quickest to emerge in BA3 which is known to receive primary input from the
thalamus.
Introduction
A goal in sensory neuroscience is to predict neuronal
responses to sensory input. Population Receptive Field (pRF) modeling has been
an important step forwards, describing the relation of various sensory inputs
with respect to neuronal responses using simple (Gaussian) functions. In the
current study, we investigate whether the pRF modeling approach can also be
applied to neuronal responses caused by vibrotactile stimulation of the
fingertips in humans as measured with functional MRI at 7 Tesla. Receptive
field modeling allows for the assessment of a somatotopic organization within
primary somatosensory cortex (S1) on the basis of receptive field centers, as
well as a categorization of receptive field size differences between sub regions
within S1 and differences in tactile stimulus input. We expect to find
somatotopic structures throughout S1 (Brodmann areas: BA1, BA2, BA3).
Furthermore, the receptive field centers are hypothesized to describe the level
of signal integration, which is expected to increase when moving up the
cortical hierarchy.Methods
Seven healthy volunteers received vibrotactile
stimulation on the fingertips (frequencies: 30-190Hz). Fingertips were
intermittently stimulated (i.e. 200ms on, 100ms off) for a period of 4 seconds.
The order of fingertip stimulation was pseudo-randomized and each digit was
stimulated 8 times per frequency. We modeled the BOLD responses according to a
Gaussian receptive field (RF) model, resulting in a preferred fingertip (i.e.
Gaussian center) and RF size (i.e. Gaussian standard deviation) per voxel. We
checked for the presence of an ordered somatotopy (i.e. spatial alignment of
preferred fingertips) and RF properties within 3 pre-selected ROIs: Brodmann area
BA3b, BA1, and BA2, which together comprise primary somatosensory cortex.
Receptive field properties were also investigated per center value and
stimulation frequency. Finally, we performed an hemodynamic response function
(HRF) estimation (32x all digits simultaneously for 0.5s) to investigate
differences in processing hierarchy between the selected Brodmann areas.Results
We found neatly ordered somatotopic structures
of finger digits throughout S1 (BA3: t(6)=18.07; p<.001; BA1: t(6)=17.83;
p<.001, and BA2: t(6)=4.53; p=.002 ) (figure 1). We additionally
found that RF sizes differed per Brodmann area (t(6)=16.86;
p<.001), showing smallest RF size for BA3 and largest RF size for BA2
(figure 2). RF size also changed per center finger digit (t(6)=3.68;
p=.005) and per applied frequency (t(6)=2.84; p=.015). These results
show that the thumb representation displays smallest RF sizes of the 5 finger
digits, just as the lowest frequency (30Hz) also displays lower RF sizes than
the selected higher frequencies. Finally, we the HRF experiment resulted in
estimated HRFs that were approximately 0.5s faster in BA3 than in BA2 (figure 3).Discussion
Using pRF modeling, we found ordered somatotopic
organizations in primary somatosensory cortical areas BA3, BA1 and BA2. Furthermore,
fingertip representations of the thumb and index finger showed smallest RF
sizes, indicating high-precision processing of somatosensory input coming from
these digits. Plausibly, the reported differences in RF size for frequency are
also caused by required specificity. The RF differences between cortical areas clearly
expose different processing stages. It is known that BA3 receives the primary
input from the thalamus, and is therefore expected to display the highest level
of detail (smallest RF sizes), whereas tactile information is believed to be
integrated within further cortical areas (larger RF sizes). These findings are
supported by the HRF experiment that showed BA3 quickest to rise, while BA2 displayed the slowest and widest curve.Conclusion
In the current study, we show that pRF modeling
can be applied to somatosensory cortex following vibrotactile stimulation in
humans measured at 7 Tesla fMRI. We were able to detect the somatotopic
organization in S1 and also describe different stages of information
integration and subsequently the inherent cortical hierarchy.Acknowledgements
This work
was supported by a grant from the National Institute of Health under Award
Number RO1MH111417.References
No reference found.