Intelligent Diagnostic System for Lung Diseases

---

Overview

Lung diseases seriously threaten human health. Reports from American Lung Association (ALA) show that more than 35 million Americans are living with chronic lung disease such as asthma, emphysema and chronic bronchitis. Every year over 349,000 Americans die from lung diseases. The lung disease death rate has been continuously increasing and become the number three killer in the United States. Lung diseases also cost the American economy $81.6 billion in the direct healthcare expenditure every year, in addition to indirect costs of $76.2 billion – a total of more than $157.8 billion. As statistics have proved that many lives can be saved and much expenditure can be reduced by early detection of lung diseases, a cost-effective, accurate and early detection tool of lung diseases is important and necessary to reduce both the risk of death and the cost of treatment resulted from lung related diseases.

Our current research interest is focused on blind source separation, feature extraction and classification techniques. Blind source separation is a powerful tool to recover the underlying source signals from their observed mixture signals. Feature extraction is used to extract parametric representation, or features, from the input signals. Classification is the last step to classify the features into different groups. Generally speaking, these three subsystems combine to form an intelligent diagnostic system.

---

Back