Lee Choonsik

PhD

National Cancer Institute

Investigator

NLST (Learn more about this study)

201206-0022

Jun 22, 2012

Investigation of the variation of the location and shape of organs using computed tomography images

The Radiation Epidemiology Branch (REB) at the National Cancer Institute (NCI) has been conducting several projects concerning radiation-induced cancer risk after the use of diagnostic and therapheutic radiation. To assess the risk of cancer risk, doses to a specific organ must be retrospectively reconstructed for the patients who were imaged or treated by radiation in the past. During the dose reconstruction, a generic organ model has been used but whether organs at risk are within or out of radiation field gives a strong impact on the resulting dose and it depends on the individual variability of organ shape and position. The physicists at the REB have investigated the variation of the organ location and shape among different invididuals using medical images and established a correlation between body dimensions (height, weight, or circumference) and the shape and position of organs. The information on the individual variability is incorporated into the dose reconstruction to understand the degree of organ dose uncertainty. We have undertaken a stomach variation study using a total of 30 adult abdominal CT images obtained from the NIH clinical center. We would like to extend the study to cover more organs such as female breast, heart, lungs, and esophagus by using the NLST image data. Once appropriate sets of CT images are obtained from the NLST database, a medical image processing tool, OsiriX, will be used to manually segment the organ contours after importing the low dose chest CT images obtained from the NLST databse. A three-dimensional organ model will be constructed and imported into another 3D image processing tool, Rhinoceros, to measure the parameters such as volume, position, dimension, center of mass, and etc. The measurements will be used to investigate correlations with body dimensions (patient height, weight, or body circumference) by using standard statistical analysis tools. From these relationships found, we will derive predictive models of organ size, shape and location based on common body mass and shape measurements.