Robustness Assessment of Automatic Reinke’s Edema Diagnosis Systems

In the past few years there has been a great interest in computer aided diagnosis research. In the field of voice quality assessment, signal processing gives us tools to analyze and extract numeric characteristics describing the analyzed signal. These features might be used to tell an impaired voice from a healthy one for many different voice conditions, being Reinke's edema one of the most severe ones. Most studies have been carried out under strict laboratory conditions, making use of professional sound equipment and facilities, trying to minimize the influence of external conditions. However, real world situations are exposed to adverse acoustic environments. The goal of this paper is to build automatic detection systems for Reinke's edema based on a novel in-house dataset and, alternatively, on the Massachusetts Eye and Ear Infirmary Voice Disorders Database, and assess noise robustness in both cases.