Fig 1.
Maximum Intensity Projections (MIPs) of corresponding in vivo-ex vivo datasets from a single mouse.
(A) In vivo MIP. (B) Ex vivo MIP. Both A and B display the vasculature with cortical depth. Unlike in the ex vivo image, where a relatively constant signal is maintained through the cortical depth, the signal in vivo is relatively weak at about 400–500 μm below the cortical surface. In addition, vessels in vivo beneath the pial vasculature (large diameter vessels at the cortical surface) are detected at a weaker signal compared to those that are not underneath these vessels (see the dark patch demarcated by the * in panel A). Scale bar = 0.2 mm.
Fig 2.
FWHM of signal along optical axis and x-axis versus depth for beads embedded in agar.
The beads were 0.5 μm diameter yellow-green fluorescent beads (excitation peak 505 nm; emission peak 515 nm) and were embedded in fructose-cleared 1% low melting point agar. Imaging was performed using 2PFM at an excitation wavelength of 800 nm. The FWHM was calculated by fitting a Gaussian to the signal profile along either the optical or x-axis for these beads. Prior to fitting the Gaussian, the image of the beads was blurred by a Gaussian with FWHM 1.5 μm, as per the vascular images on which vessel tracking was performed. Since the slope of the x-axis was not statistically different from 0 (p = 0.8136), only the PSF along the optical axis was assumed to change with depth when performing vessel tracking. The ribbons surrounding the straight lines represent the 95% confidence interval.
Fig 3.
Percent error in diameter estimation as a function of vessel diameter.
For this figure, an ex vivo image was tracked assuming a spatially varying PSF, using values from the straight line fits in Fig 2. Diameters were then recalculated for each tracked vertex assuming an unchanging PSF-width with image depth. The percent difference on this plot is the percent difference between the diameter calculated while accounting for a changing PSF, and that calculated without accounting for this spatial variance. The x-axis is the diameter calculated assuming a changing PSF (ie. the diameter initially calculated). The line displayed is an exponential fit to the data. For the small vessel diameters, where the size of the PSF is close to that of the vessel, the percent difference is appreciable. For vessels with diameters above 5 μm, this effect is much smaller (<2%).
Fig 4.
Vessel signal as a function of diameter and cortical depth.
(A) Vessel signal as a function of diameter. Signal is normalized for the ex and in vivo data by calculating the mean signal of all vessels above 10 μm diameter in each of the 4 images. The signal for each vessel is calculated separately for each image. The mean signal for vessels above 10 μm diameter is given an arbitrary value of 1, and the signal for all vessels is calculated relative to this normalized value. Smaller vessels have a weaker signal ex vivo compared to in vivo, likely due to the larger PSF ex vivo. (B) Capillary signal as a function of cortical depth. The in vivo signal is constant for the first several hundred microns, before decreasing quickly with depth (characteristic attenuation length of 171 ± 15 μm). In contrast, the ex vivo signal maintains its strength through the cortical thickness. The lines in Figs A and B are fits to the data, and the ribbons surrounding the lines are the 95% confidence intervals.
Fig 5.
The impact of vessel shadowing on capillary signal.
(A) In vivo (B) Ex vivo. The shadowing artifact is noticeably absent ex vivo (no difference in signal between shadowed/unshadowed vessels), but significant in vivo for depths below 0.6 mm.
Fig 6.
Ratio of ex vivo: in vivo vessel diameters as a function of in vivo vessel diameter.
For each vessel, the ratio of its diameter ex vivo (after correction for refractive index mismatch) to that in vivo was computed. In this figure are the ratios computed for all vessels pooled together from the four mice.