Doubling time for small pulmonary nodules is an important indicator used to diagnose lung cancer, a leading cause of death in the United States. The volume of the nodules is measured using computed tomography (CT) scans. Each volume measurement comes with a degree of uncertainty, which in turn increases the uncertainty for the doubling time measurement. Decisions regarding risky and expensive patient treatment depend on doubling time, so accuracy is important. The volume of nodules is estimated by taking a series of points marked on CT scans by radiologists and connecting these points to make a boundary. This boundary includes whole and partial pixels. By including and excluding partially filled pixels, the estimation errors can be quantified to ensure that a more accurate error estimation is made, allowing clinicians to make a better informed treatment decision. Since this process requires a radiologist to manually mark CT scans, there is a possibility for variation between radiologists, and it is time-consuming. A semi-automated method would be useful for measuring volume because it would reduce variation from radiologists' opinions and time. We can use Gaussian weighted integration to eliminate the need for radiologists to mark points on a scan. Instead, Gaussian weighted integration requires only a square boundary centered at the nodule. A Gaussian mask is applied and volume estimations are made. By simulating two scans per patient, the accuracy of each method is measured by statistical comparison with the original volume calculations, or the ground truth.
Library of Congress Subject Headings
Metastasis--Mathematical models; Lungs--Cancer--Mathematical models; Tomography--Data processing; Gaussian quadrature formulas
Department, Program, or Center
School of Mathematical Sciences (COS)
Kuhfahl, Meagan, "Emulating the volume of pulminary nodules: A Quantitative approach" (2012). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus