Various vascular mechanisms are major contributors to the pathogenesis of many ocular conditions e.g. vascular occlusion, leakage, hypoperfusion, vasospasm [
2,
13‐
15]. Therefore, different vascular imaging modalities have been adopted in ophthalmology for the management and understanding of many ocular diseases [
1]. The most commonly used modalities include dye-based angiographies i.e. FFA and ICGA, CDI, LDF, LSF and OCTA [
3]. Among these imaging modalities, OCTA is noted for its high spatial resolution, 3D tissue sampling and 2D segmentation capabilities [
16]. The introduction of SD-OCT and SS-OCT with faster scan rates and hence higher temporal resolution allowed for the development of OCTA which can construct a 3D image of the vascular plexuses within the sampled tissue based on its ability to detect areas of motion [
16]. OCTA can only indirectly provide quantitative data about velocimetric flow unlike CDI which directly measures flow velocities. Various methods are used to derive velocimetric data from OCTA e.g. plane doppler OCT, phase-based OCT, intensity-based dynamic light scattering (iDLS-OCT) and SSADA [
6]. Plane doppler OCT proved of little value in retinal imaging since the retinal blood flow is oriented nearly perpendicular to the incident OCT beam and therefore causes little doppler shift. Phase-based OCTA makes use of phase differences caused by moving particles within the sampled tissue, but is limited by a range of detectable phase differences determined by central OCT beam wavelength, tissue sampling frequency and doppler angle of the incident beam relative to the direction of motion. Therefore, velocities that create phase differences outside the predetermined range are undetectable [
6]. On the other hand, the iDLS-OCT and SSADA are insensitive to perpendicular flow and rely on recording fluctuations in the OCT signal intensity which are dependent on the rate of flow of red blood cells, acting as optical scatterers, through the sampled tissue. However, the dependency of these methods on the rate and amplitude of signal fluctuations makes them liable to the saturation phenomenon where a very low or high flow results in signal fluctuations that are outside the resolvable temporal range of the OCTA machine [
6,
8,
17]. In this study, we propose a novel mathematical method to calculate retinal and ONH flow indices that may help overcome the limitations of OCTA-based velocimetry. The proposed method profits from the high spatial resolution of OCTA by using the OCTA-measured VD as surrogate for average vessel radius. The MOPP and the VD, as a surrogate for average vessel radius, are used to calculate a flow index that is independent on velocimetric data from OCTA using the Hagen-Poiseuille flow equation which has been used in various models of the cardiovascular system physiology [
9]. The calculation of FIMs requires no secondary image processing unlike FIOs which is a significant advantage. However, these mathematical indices have their own limitations. First, they assume a fixed quantitative relation between MABP and MOAP which may not stand under certain pathological conditions e.g. carotid insufficiency. Moreover, this method assumes that perfusion pressure i.e. MOPP is the same throughout the entire vascular bed including capillaries which is not accurate since the perfusion pressure progressively decreases from larger arterioles down to capillaries. However, retinal capillaries have a certain anatomical peculiarity that makes this approximation more acceptable. Retinal capillaries lack a well-defined precapillary sphincter which is thought to be responsible for a major drop of pressure from arterioles to capillaries and therefore, retinal capillary perfusion pressure is assumed to be higher than in capillaries elsewhere in the body e.g. limb capillaries [
18]. Another disadvantage of this method is that it doesn’t differentiate between a decrease in VD secondary to vascular dropouts as opposed to a decrease in average vessel radius. Consequently, these FIMs are more suitable for longitudinal intra-patient follow-up rather than inter-patient comparison. Our results showed a positive correlation that ranged between high and moderate for the three FIMs with their corresponding FIOs. Additionally, our results showed that intraclass correlation coefficients (ICCs) were higher for the three mathematical indices FIMs than for their corresponding intensity-based indices FIOs which indicated better reliability. Moreover, ICCs were indicative of good reliability for SFIM and ONHFIM. However, ICC for DFIM was low, but still higher than DFIO. We suggest that low ICC for DFIM compared to SFIM and ONHFIM was due to the lack of statistically significant changes between baseline and follow-up values for DFIM unlike the statistically significant increase in both SFIM and ONHFIM Table
3. This was supported by the fact that despite low ICC, DFIM demonstrated the highest Pearson correlation with its respective DFIO compared to SFIM and ONHFIO Figs.
5,
6,
7, Table
2. This suggested that low ICC for DFIM reflected the lack of a statistically significant consistent change between baseline and follow-up rather than actual poor reliability. It is reported that retinal vascular autoregulation can maintain constant blood flow at a MOPP range that is 40% less or more than average normal MOPP [
19]. Since the average MOPP in our study was around 50 mmHg, we can assume a range of retinal autoregulation between 30 and 70 mmHg Table
3. We can observe in the box and whisker plots that
outliers were associated with high FIMs that were due to high MOPP exceeding the upper limit of retinal vascular autoregulation of 70 mmHg Figs.
8,
9,
10. In the box and whisker plots for SFIM, we can observe one outlier in the baseline visit and two outliers in the follow-up visit. The baseline outlier corresponded to a SFIM of 5.60 with a MOPP of 77 mmHg which exceeded the 70 mmHg limit of retinal vascular autoregulation. Similarly, the two follow-up outliers corresponded to SFIM of 6.69 and 7.42 with MOPP of 85 mmHg which also exceeded the 70 mmHg limit Fig.
8. Also, in the box and whisker plot of DFIM, one baseline and three follow-up outliers could be spotted with DFIM values of 5.91, 4,89, 5.83, 5,13 respectively with MOPP of 77, 85, 85 and 71 mmHg respectively which again all exceeded the 70 mmHg limit Fig.
9. Finally, the box and whisker plot of ONHFIM demonstrated two baseline and one follow-up outliers with ONHFIM values of 9.84, 8.40 and 9.66 respectively and concomitant MOPP of 77, 74 and 85 mmHg respectively clearly surpassing the 70 mmHg limit Fig.
10. It is noteworthy that despite these outliers for the three FIMs, none of their corresponding FIOs showed any outliers. We hypothesize that the absence of outliers for FIOs demonstrated the failure of the intensity-based SSADA to measure the high flow state resulting from MOPP exceeding the upper limit of retinal vascular autoregulation pressure of 70 mmHg due to the phenomenon of OCTA signal saturation.