Assessment of Severe Apnoea through Voice Analysis, Automatic Speech, and Speaker Recognition Techniques

This study is part of an ongoing collaborative effort between the medical and the signal processing communities to promote research on applying standard Automatic Speech Recognition (ASR) techniques for the automatic diagnosis of patients with severe obstructive sleep apnoea (OSA). Early detection of severe apnoea cases is important so that patients can receive early treatment. Effective ASR-based detection could dramatically cut medical testing time. Working with a carefully designed speech database of healthy and apnoea subjects, we describe an acoustic search for distinctive apnoea voice characteristics. We also study abnormal nasalization in OSA patients by modelling vowels in nasal and nonnasal phonetic contexts using Gaussian Mixture Model (GMM) pattern recognition on speech spectra. Finally, we present experimental findings regarding the discriminative power of GMMs applied to severe apnoea detection. We have achieved an 81% correct classification rate, which is very promising and underpins the interest in this line of inquiry.

Quantitative electroencephalography reveals different physiological profiles between benign and remitting-relapsing multiple sclerosis patients

BACKGROUND A possible method of finding physiological markers of multiple sclerosis (MS) is the application of EEG quantification (QEEG) of brain activity when the subject is stressed by the demands of a cognitive task. In particular, modulations of the spectral content that take place in the EEG of patients with multiple sclerosis remitting-relapsing (RRMS) and benign multiple sclerosis (BMS) during a visuo-spatial task need to be observed. METHODS The sample consisted of 19 patients with RRMS, 10 with BMS, and 21 control subjects. All patients were free of medication and had not relapsed within the last month. The power spectral density (PSD) of different EEG bands was calculated by Fast-Fourier-Transformation (FFT), those analysed being delta, theta, alpha, beta and gamma. Z-transformation was performed to observe individual profiles in each experimental group for spectral modulations. Lastly, correlation analyses was performed between QEEG values and other variables from participants in the study (age, EDSS, years of evolution and cognitive performance). RESULTS Nearly half (42%) the RRMS patients showed a statistically significant increase of two or more standard deviations (SD) compared to the control mean value for the beta-2 and gamma bands (F = 2.074, p = 0.004). These alterations were localized to the anterior regions of the right hemisphere, and bilaterally to the posterior areas of the scalp. None of the BMS patients or control subjects had values outside the range of +/- 2 SD. There were no significant correlations between these values and the other variables analysed (age, EDSS, years of evolution or behavioural performance). CONCLUSION During the attentional processing, changes in the high EEG spectrum (beta-2 and gamma) in MS patients exhibit physiological alterations that are not normally detected by spontaneous EEG analysis. The different spectral pattern between pathological and controls groups could represent specific changes for the RRMS patients, indicative of compensatory mechanisms or cortical excitatory states representative of some phases during the RRMS course that are not present in the BMS group.