Automated Extraction and Diagnosis of Lung Emphysema from Lung CT Images Using Artificial Neural Network

Abstract

Emphysema is characterized by loss of elasticity of the lung tissue; destruction of structures supporting the alveoli; the destruction of capillaries feeding the alveoli. The result is that the small airways collapse during expiration, leading to an obstructive form of lung disease (air is trapped in the lungs in obstructive lung diseases). The scientific definition of emphysema is: "permanent destructive enlargement of the airspaces distal to the terminal bronchioles without obvious fibrosis". Hence, the definite diagnosis is made by a pathologist. At present, diagnosis of emphysema is done by using spirometry, X-rays, spiral chest CT-scan, bronchoscopy, blood tests, pulse oximetry and arterial blood gas sampling. Although emphysema is an irreversible degenerative condition, early prognosis and treatment are very important for optimizing the patients' quality of life. This paper proposes an automated computed-aided diagnosis algorithm for extracting enlarged airways from lung CT image automatically using an image matching method, and consequently classifying the lung condition artificial neural network (ANN) by supplying 30 network inputs obtained from texture analysis of the lung CT image and calculations of the feature properties of extracted enlarged airways to the trained ANN. Our research aims to produce an automated system which has higher objectivity in the diagnosis of lung emphysema