Audio Technology To Detect Lung Cancer Earlier

In the UK lung cancer is the leading cause of cancer death, also, UK survival rates are poorer compared to other European countries. Lung cancer is plagued by late presentation; 70% present with advanced incurable disease and a third die within 90 days of diagnosis. As such, there is a clear and urgent need to achieve earlier diagnosis of lung cancer. Symptomatic presentation is the most common route to lung cancer diagnosis and symptoms may be present for many months before diagnosis, even in early stage disease. The most common (68%) presenting symptom is cough. Unfortunately cough is also common with other illnesses. Furthermore, a high proportion of those at high-risk of lung cancer have pre-existing cough or respiratory disease (e.g. ex- or current- smokers, patients with chronic obstructive pulmonary disease (COPD)). Cough sounds are known to vary according to underlying lung pathology and could therefore have diagnostic value. Potentially, there may be unique cough and/or respiratory sounds or patterns associated with lung cancer that are not detectable by the human audio spectrum. Identification of a tool that accurately discriminates lung cancer cough could be pivotal.

This is a prospective cross-sectional study that will involve subjective analysis of spectrograms of cough recorded from individuals with "normal" lungs, individuals at high-risk for lung cancer (COPD and other chronic lung diseases) and individuals with lung cancer. 24 hour ambulatory audio recordings will be prospectively collected from patients attending respiratory medicine clinics at Queen Elizabeth University Hospital (QEUH), Glasgow. Participants will be given a free-field lapel microphone and mp3 recorder for 24h. The Leicester Cough Monitor (LCM) will be used to extract the cough sounds from the 24h recordings. The LCM is an automated cough detection system that was developed by Dr Surinder Birring (Kings College Hospital). It uses an algorithm to automatically identify cough sounds from audio recordings, which is then able to provide data on cough frequency. As part of this process, the LCM splices out 1 second sound clips for all parts of the audio recordings that are identified as being (i) a cough sound or (ii) a non-cough sound.