Focal infrared neural stimulation with high-field functional MRI: A rapid way to map mesoscale brain connectomes

Abstract

We have developed a way to map brain-wide networks using focal pulsed infrared neural stimulation in ultrahigh-field magnetic resonance imaging (MRI). The patterns of connections revealed are similar to those of connections previously mapped with anatomical tract tracing methods. These include connections between cortex and subcortical locations and long-range cortico-cortical connections. Studies of local cortical connections reveal columnar-sized laminar activation, consistent with feed-forward and feedback projection signatures. This method is broadly applicable and can be applied to multiple areas of the brain in different species and across different MRI platforms. Systematic point-by-point application of this method may lead to fundamental advances in our understanding of brain connectomes.

Figures

Fig. 1. Global connections revealed by INS stimulation in cat visual cortex.

(A) Stimulation paradigm. Each block of INS comprises 20 trials of pulse trains at a single energy. Train duration, 0.5 s; intertrain interval, 35.5 s; pulses, 250 μs; 200 Hz; 0.1 to 1.0 J/cm2 per pulse. (B) Connections of cat area 17/18. Red arrows indicate cat visual projections from area 17 [adapted from (49)]. The color bar shows T statistics for significant voxels in (C) and (D), thresholded at P < 0.01. (C to E) Data from cat #4. (C) Thalamic activations. Left: Stimulation of area 17/18 at 0.3 J/cm2 [red asterisk in (D)] reveals ipsilateral LGN activation. Right: At 0.7 J/cm2, stimulation produces larger ipsilateral activation in LGN and pulvinar (LGN/Pul). In addition, strong activations are seen in contralateral pulvinar and lateral posterior nucleus (LP). (D) Cortical activations in three slices (dorsal, intermediate, and ventral). The red asterisk denotes the stimulation of area 17/18 at 0.7 J/cm2. Significant voxels seen in areas 17, 18, and 19; AMLS; anterior ectosylvian area (AES); and cingulate visual area (CVA). (E) Intensity dependence (average of 20 trials; error bars, SEM). Top: Time courses (left) and amplitudes (right) at the laser tip site in (D). The yellow bar indicates INS onset. Bottom: Response amplitudes at connected ipsilateral 18 and contralateral 17/18 sites in (D).

Fig. 2. All slices of voxels activated by laser stimulation of the area 17/18 border (0.7 J/cm2).

Significant voxels (P < 0.01, FDR of 20%) overlie locations with known anatomical connections to laser stimulation site in area 17/18. These include LGN; LP/pulvinar; areas 17, 18, 19, 20, and 21; and sylvian areas (AMLS and PMLS). Polysynaptic areas include contralateral LP/pulvinar, AES, CVA, and areas 4, 5, and 7 (see the main text). Note that LGN, which has very strong connections with area 17/18, contains some of the most significant voxels. The color bar shows T statistics for significant voxels. See Fig. 1A for the expanded forms of the abbreviations.

Fig. 3. Local connections revealed by INS stimulation in squirrel monkey somatosensory cortex.

(A) Somatotopy in areas 3b and 1 (38). (B) Stimulation paradigm. ISI, inter-stimulus interval. (C) A 400-μm optic fiber on top of an artificial dura in an optical chamber over SI. Surface coil, 3 cm. Dark vessel, central sulcus. (D) Two planes of MRI imaging, with the chamber seen on top of the cortex. Scale bar, 1 cm. (E) Schematic of digit connectivity in SI (33). (F) Schematic of feed-forward and feedback inter-areal connectivity in SI (35). (G) Cortical responses to stimulation at D3. Slice in tangential plane. The red dot indicates the optical fiber tip in area 3b. See Fig. 4 (A and C) for determination of areal borders (dashed white lines). CS, central sulcus; M, medial; P, posterior. Scale bar, 5 mm (monkey #2). (H) Cortical response to stimulation of area 2 (red arrowhead, fiber tip; dark spot, signal dropout). Slice in orthogonal plane. Green arrowheads indicate feed-forward activations. White and blue arrowheads denote feedback. See Fig. 4 (B and D) for determination of areal boundaries. The white rectangle indicates the silicone plug in the chamber. D, dorsal; LS, lateral sulcus. Scale bar, 5 mm. The color bar shows T statistics for significant voxels in (G) and (H) (monkey #3). (Photo credit: Gang Chen and Anna Wang Roe.)

Fig. 4. Determination of areal boundaries via optical imaging

(A) Monkey #2 (Fig. 3G, tangential case) and (B) monkey #3 (Fig. 3H, orthogonal case). Left: Optical chamber. Blue boxes denote imaged fields of view. Right: Optical images to stimulation of digit tips [cf. (42)]. White arrows in D3 of (A) correspond to those in (C) and (E). Field of view for D1 and D2 images is shifted relative to that for D3 and D4. (C and D) Summary of digit tip topography from (A) and (B). (C) Arrows indicate corresponding locations with (E). The red dot indicates the laser stimulation site. White dotted lines denote the borders between areas. (D) Colored arrowheads indicate corresponding locations with (F). The red arrowhead denotes the laser stimulation site. The horizontal purple line indicates the approximate location of slice in (F). Green arrowheads denote the feed-forward projections in (F). White (area 3b) and blue (area 1) arrowheads indicate feedback projections in (F). (A and C) Monkey #2. (B and D) Monkey #3. (E and F) Compare with (C) and (D) for alignment of optical imaging and fMRI maps. (E) Same as Fig. 3G. (F) Same as Fig. 3H. These figures are shown to easily see the correspondence between optical and MRI images. (Photo credit: Robert M. Friedman and Anna Wang Roe.)

Fig. 5. Intensity dependence.

(A) Time course of MR signal change at different intensities of INS stimulation (average of 13 trials). The same case as Figs. 3G and 4 (A, C, and E) (monkey #2). (B) Each dot denotes the peak amplitude of MR signal of individual trials (n = 13). Error bars, SEM. The dashed line represents the regression line. (C and D) Local activation at high [(C) 0.50 J/cm2] and low [(D) 0.29 J/cm2] intensities. The same case as in Figs. 3H and 4 (B and D) (monkey #3). Stimulation in area 2 (optical fiber, red arrow). Third panel: Same as in Fig. 3H. Adjacent 1.5-mm-thick slices from lateral (left) to medial (right). Central sulcus [seen in the two rightmost panels of (C)] aided in demarcation of area 3b (white arrowheads) and area 1 (blue arrowheads). (D) Same as (C) except for lower INS intensity. Scale bar, 5 mm. The color bar shows T statistics for significant voxels in (C) and (D).

Fig. 6. Reliability of activation.

Panels A, B, C and D represent adjacent slices. Top row: Original panels (Fig. 5 C, n = 13 trials) showing feed-forward (green arrowheads) and feedback (white and blue arrowheads) connections from point stimulation in area 2 (red arrow). Middle row: Sum of even trials (n = 6). Bottom row: Sum of odd trials (n = 7). Since half as many trials will have lower statistical power, we used P < 0.0005 for half-trial images (middle and bottom rows) instead of the P < 0.0001 used for all trials (top row). Comparison of the half-trials reveals that the images are largely similar (from monkey #3).