Model-Based Image Reconstruction for X-ray CT in Lung Imaging Modern X-ray computed tomography (CT) systems provide high-quality images for diagnosing numerous conditions including a variety of lung diseases. Unfortunately, technological advances in CT imaging have been accompanied by significant increases in X-ray radiation dose to patients. There is growing concern about the public health consequences of such doses. Furthermore, even with typical levels of radiation dose, current X-ray CT images have suboptimal image quality due to the limitations of the traditional image reconstruction algorithms used in clinical systems. We propose to develop, implement, analyze and evaluate model-based image reconstruction (MBIR) methods for X-ray CT to improve image quality in lung imaging and to reduce patient dose. Unlike commercially available denoising methods, the proposed MBIR methods are based on accurate models for the physics and statistics of X-ray CT systems. The methods will use edge-preserving regularization that is tailored to lung scans to control noise while improving spatial resolution. We will develop techniques for accelerating the iterative algorithms used in MBIR methods. The methods will be evaluated using computer simulations, phantom studies, and human studies. Specifically, we will focus here on lung CT applications, including morphological characterization of lung nodules and assessment of pulmonary diseases. The clinical impact of MBIR methods will be studied using automated lung image analysis tools and radiologist observer studies.
The relevance of this research to public health is that we will develop and evaluate sophisticated techniques for processing the raw data measured by X-ray CT scanners to dramatically reduce the X-ray radiation dose to patients and to further improve the image quality in lung CT imaging for more accurate diagnosis and treatment.
Showing the most recent 10 out of 11 publications
