Image analysis provides powerful tools for extracting meaningful and actionable information from varied image modalities. Applications of image analysis are incredibly diverse across multidisciplinary sciences. This dissertation investigates three applications of image analysis that impact human health on a global scale. First, the quantification of point-source emission plumes of greenhouse gases through image analysis of spectrometer data is improved using a novel processing algorithm and improved detection criteria. The developed algorithms improve processing speed, enabling faster delivery of actionable information from collected images and improving the accuracy of carbon dioxide concentration estimates. Second, effective radiation therapy for lung cancers relies on understanding the tumor motion. Image analysis of computed tomography volumes collected before treatment allows learning patient-specific respiratory motion patterns to provide more accurate tumor targeting. These motion tracking algorithms improve the accuracy and latency of treatment. Third, a scaleable method for dynamic radiotherapy dose tracking is investigated, initially for head and neck cancer patients. Head and neck cancer patients are especially prone to weight loss and a dynamic soft tissue presentation. A rigid consideration of tumor and surrounding organs limits the accuracy and applicability of dose tracking. Deformable image registration enables a dose tracking method to demonstrate the consideration of dynamic soft tissue presentation for these patients. These applications improve the image analysis capabilities and availability of actionable information for climate scientists and cancer care clinicians. Image analysis within this dissertation extracts valuable insights from images across domains to address global-scale issues directly impacting world health.
Posted by: Nathan Galli