Towards intelligent classification and staging of acoustic voice data in cancer of the larynx patients
R. T. Ritchings · 2002
This work describes a investigation into the feasibility of using signal processing techniques, followed by intelligent classification to derive an objective, graded assessment of a patient's voice quality following treatment for cancer of the larynx. At present in the UK assessment and rehabilitation planning is performed by Speech and Language Therapists (SALT) who endeavour to rehabilitate a patient's voice back to normality, or as near normal as possible, quickly following treatment. Earlier work by Ritchings et al. (1999, IEEE SMC99, 2, 340-345) has demonstrated some success when using Artificial Neural Networks (ANN) in conjunction with short-term and long-term frequency-domain parameters taken from electrical impedance (EGG) signals recorded from larynx cancer patients at different stages of recovery. The signal processing techniques used to derive these parameters, and the ANN that was used for the classification, are reviewed here, and the progress that has been made when applying these techniques to a patient's acoustic signals is described. The project aim is to develop a system that will perform real-time processing and classification in the SALT's clinic, with the result superimposed visually on the patient's signal waveform. (4 pages)