The development and proliferation of quantitative image analysis methods have accelerated research efforts and are having an increasingly significant impact in modern clinical practice. Although the research utility of these techniques has been amply demonstrated in determining longitudinal and groupwise trends, they are also becoming increasingly relevant in the clinical setting in providing biomarkers for aiding patient diagnoses, monitoring disease progression, and determining treatment outcomes. Increases in the capabilities and accessibility of computational facilities and a corresponding sophistication in computational algorithms have only made such practices more commonplace. One of the most significant hurdles for the pulmonary imaging field in adopting more quantitative clinical practices and exploring additional novel research pathways is the open availability of accurate, robust, and easy-to-use image analysis tools. This project addresses this important unmet need by bringing together leading expertise in pulmonary image analysis at the University of Pennsylvania, the University of Virginia, and the University of Iowa to develop, evaluate, and deploy a critical open-science resource under community support that would allow access to multi-modality data and tools for processing and analysis of lung imaging data. Toward this end, a comprehensive image analysis and data package, denoted as ITK-Lung, will be developed to address the specific needs of the multi-modality pulmonary imaging community. This first-of-its-kind resource will be enhanced by data- and domain-specific tuning that will accommodate a variety of user backgrounds and needs. These developments will be evaluated through their application to real-world use cases representing the state-of-the-science in lung imaging research. The successful completion of this project will help to fully realize the value of computational image analysis in pulmonary research and lead to new insight that may open novel ways to treat, cure and even prevent pulmonary disorders.
This project will improve pulmonary scientists' ability to explore biological and clinical hypotheses concerning the structure and function of the human lung using multi-modality imaging data. Scientific research has been significantly enhanced by recent emphases on open-data and open-source tools, yet no such publicly available, general computational resources exist for pulmonary imaging. By providing openly accessible, user-friendly, widely interoperable, and extensively validated tools for multi-modality pulmonary imaging analysis and processing, the project will enable a broad field of scientists to leverage modern imaging technologies more effectively in answering basic science questions about the lung that will in turn lead to new clinical insights and advancements.
| Tustison, Nicholas J; Avants, Brian B; Lin, Zixuan et al. (2018) Convolutional Neural Networks with Template-Based Data Augmentation for Functional Lung Image Quantification. Acad Radiol : |
| Qing, Kun; Tustison, Nicholas J; Mugler 3rd, John P et al. (2018) Probing Changes in Lung Physiology in COPD Using CT, Perfusion MRI, and Hyperpolarized Xenon-129 MRI. Acad Radiol : |
