Abstract
Background and Aims
Vascular access dysfunction is one of the leading causes of morbidity and a major contributor to healthcare costs in hemodialysis (HD) patients. Inexpensive, non-invasive tools for routine assessment of vascular access function are needed. Hemodynamically relevant stenoses in arteriovenous fistulas (AVF) lead to a reduction in access flow rate (Qa) and changes in blood flow patterns in the AVF that may be picked up by palpation and auscultation. We hypothesized that these changes in blood flow patterns can not only be felt and heard but also seen, i.e., that they may be detectable in video recordings done with commercially available smartphones after digital motion augmentation.
Methods
We studied HD patients with AVF dysfunction requiring balloon angioplasty and/or stenting. One-minute video recordings of the skin above the AVF and Qa measurements were conducted before and after the endovascular intervention. Videos were recorded with an iPhone 6S (Apple Inc., Cupertino, CA, USA). Qa was measured by HVT100 endovascular flowmeter (Transonic Systems Inc., Ithaca, NY, USA). Significant access stenosis was defined as a >50% reduction of luminal diameter. Degree of stenosis was assessed by angiography. Frame-to-frame pixel changes in video images were amplified using “Eulerian Video Magnification” (Massachusetts Institute of Technology, MA, USA; http://people.csail.mit.edu/mrub/evm/#code). The time-domain data were then transformed into the frequency-domain signals. Fifty random 10-second segments were sampled per one-minute video, and the frequency with the lowest magnitude (Fmin) was determined in each sample (example shown in Fig. 1). The average Fmin was then assessed for its association with the degree of AVF stenosis.
Results
Ninety subjects were studied (Table 1). AVF interventions were successful in all patients. Post-intervention Qa (1638 ± 714 ml/min) was on average 1.23-fold higher than pre-intervention Qa (1373 ± 684 ml/min; P<0.01, paired t-test). Subjects were grouped by degree of stenosis, and the number of subjects in each category is shown in Fig. 1B. Higher degrees of stenosis were associated with greater increases in Qa from before to after the intervention (P<0.01, one-way ANOVA; Fig. 1C). Interestingly, the degree of AVF stenosis was also positively related with the change in Fmin from before to after the intervention (P=0.08, one-way ANOVA; Fig. 1D).
Conclusion
Simple smartphone video recordings of AVF appear to contain frequency-domain information that correlates with hemodynamic changes caused by AVF stenoses. While the Fmin metric employed in our analysis is not ideal, these results should encourage the quest for other parameters that exhibit higher correlations with vascular access dysfunction. If successful, this would allow commercially-available smartphones to be used as ubiquitous tools for quick, non-invasive, ambulatory surveillance of AVF function, thereby allowing timely referrals and avoidance of emergency interventions.
Correlation of xenon-enhanced computed tomography-defined cerebral blood flow reactivity and collateral flow patterns.
