Purpose:
To detect and localize retinal parameters that change according to the cardiac cycle.
Methods:
Video sequences of the ocular fundus has been digitized online (Zeiss fundus camera, green filter, CCD camera, 25 fps, 768 x 494 pixels). Sequences were aligned offline to compensate for eye movements (AutoVisualize, AutoQuand). For further evaluation, stacks of 128 aligned images (5,12 s) with Δt = 40 ms temporal spacing were used (time domain stack TDS). For any selected position (xn,ym), a z–profile F(xn,ym,t) can be calculated that shows the time course of the intensity at this position. All 768 x 494 functions F(xn,ym,t) were Fourier transformed (Matlab). The absolute values of the coefficients of the resulting functions F’(xn,ym,f) were used to create a new stack of 64 images, each corresponding to a frequency between 0 and 12,5 Hz with Δf = 0,1953 Hz spacing (frequency domain stack FDS). Each image of the FDS shows the spatial distribution of the amplitude of the Fourier coefficient for the selected frequency. A z–profile shows the frequency spectrum at the selected position (xn,ym).
Results:
The time course of fundus reflection (z–profile in TDS) shows a pulsatile component according to the cardiac cycle. The reflection decreases during systole (increasing blood volume) and increases during diastole (decreasing blood volume). At the optic disk the reflection changes by about 3% (mean value over the area of the optic disk). The period TRR of the cardiac cycle can be calculated from F(x,y,t) (Tornow RP, et al. IOVS 2003;44:ARVO E–Abstract 1296). The spectrum (z–profile in FDS) shows a clear peak at f = 1/TRR. No other pronounced peak is present. The Fig. shows the spatial distribution of the Fourier coefficients of f = 1/TRR. At the position of the optic disk, this amplitude is high. Areas outside the optic disk show much lower or no amplitude for this coefficient except for some areas close to the border of large arteries (→). This is due to small movements of vessel segments according to the cardiac cycle.
Conclusions:
Besides areas with pulsatile change in reflection, the position and amplitude of moving vessel segments can be easily identified. This might be useful to analyze the change of vessel parameters (e.g. compliance) over time.
Keywords: blood supply • optic disc • imaging/image analysis: non-clinical