Skip to main content

more options

Automated Emphysema Quantification

Brad Keller (Alumnus)


Emphysema is clinically defined as abnormal enlargement of the air spaces in the lung. It is part of a larger class of disease known as Chronic Obstructive Pulmonary Disease (COPD). According to the National Emphysema Foundation

Emphysema is the fourth largest cause of mortality in the U.S., yet it is not a highly publicized disease, consequently, little is known about it by the average citizen. Almost 16-30 million U.S. citizens are afflicted with emphysema... and every year approximately 100,000 sufferers die of the disease. Smoking is the major cause, but with ever increasing air pollution and other environmental factors that negatively affect pulmonary patients, those numbers are on the rise.

Standard automated emphysema quantification from Computed Tomography images involves thresholding at a certain H.U. density level, and taking the percentage of lung parenchyma below that density. This defines the Emphysema/Pixel Index which is the most common measure of emphysema in patients from CT images. The Emphysema Index, however, has several disadvantages in practice. First is its poor correllation to PFT and DLCO scores. Second, it has very low reproducibility between scans, leading to wide variations in scores. Third, it is very dependant on noise and inspiration levels.

Some of our approaches to solving these problems involve personalizing the emphysema threshold on a scan-to-scan basis based on noise and tracheal air. Another aspect we are currently researching is compensation for the variation in inspiration level over sequential scans. We are also evaluating characterizing the percentage of healthy lung parenchyma, instead of a standard emphysema index score.

List of Current Research Projects