Abstract
With the advances of electronics and software technologies in the last decade, an important new direction in sonotubometry has been created concerning an assessment of the Eustachian tube (ET) function under physiological conditions. Despite the fact that the sonotubometry technique has gradual been improved in the last twenty five years, it is not yet used systematically to evaluate Eustachian tube ventilatory function, because its reproducibility, history, mapping to the context (patient, clinic data, medication, a.o.) validation and value for clinical practice have not yet been consolidated and integrated in a service oriented, knowledge-based system that nowadays can use advanced modelling, segmentation, search, data analytics and prediction tools. The utilisation of Digital Signal Processing techniques for sound acquisition and low-level processing, noise rejection, attenuation monitoring and digital feature-based representation of the sound records create the premises to confirm the validity and reproducibility of sonotubometry as test method. This is one of the primary contributions of the research reported in this paper. Another objective is the integration of the components necessary to build a knowledge-based system used in context-driven Eustachian tube function evaluation and decision support for otological diagnostic: feature-based modelling of the digitized sound records; context-driven mapping of sound records in a knowledge base with multiple tagging; iterative learning process; data analytics; predictive analysis for decision optimization. The pilot implementation of the system and experiments are provided.