The broad, long term objective of the proposed research is to develop a fully automated, computerized system that will assist radiologists in the quantitative assessment of pleural-based diseases in helical computed tomography (CT) scans. The objective disease quantification that such a system would produce is expected to provide valuable information regarding disease progression or the response of disease to treatment. This information would be used to guide patient treatment or gauge the efficacy of therapeutic agents during clinical trials. This specific study will focus on one particular pleural-based disease, malignant pleural mesothelioma. Radiologists must inspect the anatomic region between the lungs and the chest wall or the mediastinum on CT scans to evaluate the extent of mesothelioma, which presents as a thickening of the pleura. Currently, assessment of mesothelioma severity is based on manual measurements of the pleural thickness made at several image locations; this approach represents a coarse, tedious process that suffers from subjectivity and variability among observers who make the measurements. The development of automated and semi- automated methods that, in each hemithorax of a CT scan, delineate the lung border, delineate the chest wall and mediastinum, and quantify the differences between these anatomical structures as a measure of pleural thickness, is expected to facilitate the efficient, objective assessment of pleural thickness for the evaluation of mesothelioma tumor. The utility of such consistent methods is accentuated by their application to temporally sequential scans, in which a comparison of tumor size over time is made to determine the rate of disease progression or the response of disease to a specific treatment regimen. The hypothesis of this study is that computerized techniques can be developed to assist radiologists and clinicians in the reliable, consistent, and reproducible quantification of mesothelioma tumor as an objective parameter for initial staging and subsequent treatment response.
Showing the most recent 10 out of 12 publications
