Analysis suggests that daily self-imaging generated images of satisfactory quality for human grading and automated analysis of fluid volume over time permitting a novel paradigm of monitoring. Home OCT may allow personalized retreatment decisions with fewer clinic visits especially in conjunction with longer acting drugs and sustained release systems.
To evaluate the longitudinal performance of the Home OCT System comprising an SD-OCT device for self-imaging at home, telemedicine infrastructure for automated data upload, and deep learning algorithm (NOA) for fluid quantification. The subjects were enrolled for a median of 73.5 days, self-imaged on average in 49 of these days, and on some days more than once, for a total of 810 self-images. Of these, 93% had satisfactory quality. The median (IQR) self-image duration was 49 seconds. For any fluid presence, the NOA agreed with human grading in 94% of cases, 99% for IRF and 91% for SRF. Cases of fluid falsely identified by NOA were limited on average to very small volume of less than 1 nL of any fluid type.
