Optical Methods for Tympanic Membrane Characterisation : Towards Objective Otoscopy in Otitis Media

Abstract: Otitis media, which is an upper respiratory tract infection that affect the middle ear, is the second most common disease in childhood, outnumbered in prevalence only by the common cold. Diagnosis of middle ear inflammation is often performed in the primary healthcare where the normal procedure involves anamnesis and physical examination of the tympanic membranes (TM) of the patient, usually be means of otoscopy. The general aim of this thesis was to develop optical methods that enable quantification of TM characteristics associated with otitis media. Diffuse reflectance spectroscopy was applied to quantify TM erythema using previously suggested erythema detection algorithms. Healthy TM:s were significantly distinguished from TM:s with induced erythema (p < 0.01) and from TM:s in ears with mucous middle ear effusion (p < 0.05). A new technique for surface shape assessment based on an on-axis dual fibre array incorporated in an otoscope was developed and evaluated in ear models and on tympanic membranes from harvested temporal bones. The technique utilises the combined effects of source-detector fibre separation and fibre-to-sample distance on the detected light intensity.Optical phantoms, both polyacetal plastic solids and latex membranes, were utilised to demonstrate the ability of the surface shape assessment technique to differentiate between convex and concave surfaces – as a bulging tympanic membrane is typically associated with acute otitis media whereas a retracted eardrum is associated with otitis media with effusion. Monte Carlo simulations of the surface shape data were performed in order to validate the experimental results with a theoretical model that are consistent with light transport theory. Retracted and bulging tympanic membranes from harvested temporal bones could be separated with a single measurement, given that variations in measurement distance were accounted for and that measurement from normally positioned tympanic membranes were used for signal normalization. In conclusion, the studies implicate that for individual otitis diagnosis, the hyperaemic tympanic membrane was separated from the healthy by application of erythema indices using diffuse reflectance spectroscopy. Moreover, bulging and retracted positions of the tympanic membrane were separable by means of the source-detector intensity matrix. For further clinical studies it is reasonable to assume that data from both methods are needed for diagnosis.