Research Interests

• Lung and lung lobe segmentation: We are developing algorithms for automatic segmentation of the lung and lung lobes in CT images of humans and animals.
• Airway segmentation: We have an automatic algorithm for identifying the 3D human airway tree in CT images.
• Endoscopic image analysis: We are analyzing the color of the airways as viewed through a bronchoscope and then trying to map these colors onto the segmented airway tree obtained from a 3D CT scan.
• Inter- and intra-subject lung registration: To match lung anatomy across time and across individuals, we are developing lung registration methods for inter-subject and intra-subject matching.  We are using the inter-subject matching to develop a computerized lung "atlas" that contains the major anatomic structures of the lung.
• Regional lung expansion and parenchymal strain: With to lung CT images taken at different inflation levels, we estimate regional lung expansion and parenchymal strain by tracking anatomic structures across images.   We hope to use these measures as indicators of early lung disease.
• CAD: Geometric and grayscale properties can be used to detect and classify abnormalities.  There are ongoing projects applying CAD to lung diseases such as emphysema and cancer, and to breast cancer.