A novel, automated measurement of spontaneous vein pulsation from infrared videos of the optic disc.

Abstract

Purpose : Reduced or absent spontaneous retinal venous pulsation (SVP) is a sign of Intracranial Hypertension (IH). SVP can be assessed subjectively through fundoscopy. We introduce an automated and objective measurement of SVP using infrared video recording of the Optic Disc (OD).

Methods : We recruited 10 healthy volunteers and analysed a 20 second infrared video recording of their OD using a Spectralis OCT-1 (Heidelberg Engineering, Germany). During acquisition, each volunteer was asked to breathe normally for 10 seconds and then perform a Valsalva manoeuvre for 10 more seconds in an attempt to voluntarily increase vein pressure and impair the SVP. A selected area around a vein at the OD from each video was then processed using a novel algorithm. In short, the algorithm stabilised the video around the vein; filtered it in the space domain to isolate structures of the diameter size of retinal vasculature; filtered it in the time domain to isolate changes within the range of the human heart-rate; calculated frame-by-frame differences of the real part of the Fourier transform to produce a trace of pulsation. The peak-frequency amplitude of each trace was calculated to check whether a signal-reduction was detected during the manoeuvre. Differences between the two sequences of the video were calculated using a generalised linear model and random effects to account for repeated measurements (resting state [10 s.] and Valsalva manoeuvre [10s.]).

Results : A significant amplitude reduction in the SVP trace was measured during the manoeuvre (p < 0.0001), with the amplitude reduced by a factor of 0.57 [95% confidence interval: 0.45, 0.67] from the resting state.

Conclusions : This work introduces a novel objective method for the measurement of SVP that can identify different pulsation amplitudes within the same subject. Future work will test the measurement in patients with known ICP to assess its ability to quantify SVP across subjects.

This is a 2020 ARVO Annual Meeting abstract.

 

The 10 traces calculated from the10 videos of the same number of volunteers.