Otitis media (OM) is a common ear infection and a leading cause of conductive hearing loss in the pediatric population. Current technologies can reasonably diagnose the infection with a sensitivity and specificity of 50–90% and 60–90%, respectively. However, these techniques provide limited information about the presence of biofilm or fluid formed behind the tympanic membrane (TM). Our group has developed handheld probes and portable optical coherence tomography (OCT) systems that have been used in various clinical studies to provide quantitative information about structural changes, and thus accurately characterize OM. Further, an automated machine learning-based approach from our group has been developed and integrated to classify OCT images associated with various stages of OM, without the need for interpretation by an expert reader.
In this study, we report a portable, low-cost, briefcase OCT system with automated classification for point-of-care diagnosis of OM. The briefcase OCT system cost < $8000USD with a 5-fold cost reduction and a 3-fold size reduction, compared to more standard OCT systems. Additionally, this system utilizes unique real-time mosaicking of surface video images that are synchronized with rapid A-scan acquisition, enabling computationally generated thickness maps and construction of cross-sectional B-mode images over extended lateral distances. Furthermore, a random-forest based classifier is utilized with an expanded feature set based on various statistics and metrics derived from OCT A-lines and B-scans. This system will help physicians and untrained users to collect OCT data and receive a diagnostic prediction indicating the presence and type of OM, potentially leading to more accurate point-of-care diagnoses.
|